Skip to main content

Full text of "Bulletin of the Natural History Museum Zoology"

See other formats


ISSN 0968-0470 



Bulletin of 

The Natural History 

Museum 



Zoology Series 




THE 

NATURAL 
HISTORY 
MUSEUM 



VOLUME 68 NUMBER 2 28 NOVEMBER 2002 



The Bulletin of The Natural History Museum (formerly: Bulletin of the British Museum 
(Natural History) ), instituted in 1949, is issued in four scientific series, Botany, 
Entomology, Geology (incorporating Mineralogy) and Zoology. 

The Zoology Series is edited in the Museum's Department of Zoology 
Keeper of Zoology Prof P.S. Rainbow 

Editor of Bulletin: Dr BT. Clarke 

Papers in the Bulletin are primarily the results of research carried out on the unique and ever-growing collections of the 
Museum, both by the scientific staff and by specialists from elsewhere who make use of the Museum's resources. Many of the 
papers are works of reference that will remain indispensable for years to come. All papers submitted for publication are 
subjected to external peer review before acceptance. 

SUBSCRIPTIONS Bulletin of the Natural History Museum, Zoology Series (ISSN 0968-0470) is published twice a year 
(one volume per annum) in June and November. Volume 68 will appear in 2002. The 2002 subscription price (excluding VAT) 
of a volume, which includes print and electronic access, is £88.00 (US $155.00 in USA, Canada and Mexico). The electronic- 
only price available to institutional subscribers is £79.00 (US $140.00 in USA, Canada and Mexico). 

ORDERS Orders, which must be accompanied by payment, may be sent to any bookseller, subscription agent or direct to the 
publisher: Cambridge University Press, The Edinburgh Building, Shaftesbury Road, Cambridge CB2 2RU, UK; or in the 
USA, Canada and Mexico: Cambridge University Press, Journals Department, 40 West 20th Street, New York, NY 101 1-421 1, 
USA. EU subscribers (outside the UK) who are not registered for VAT should add VAT at their country's rate. VAT registered 
members should provide their VAT registration number. Japanese prices for institutions (including ASP delivery) are available 
from Kinokuniya Company Ltd, P.O. Box 55, Chitose, Tokyo 156, Japan. Prices include delivery by air. Postmaster: send 
address changes in USA, Canada and Mexico to: Bulletin of the Natural History Museum. Zoology Series, Cambridge University 
Press, 110 Midland Avenue, Port Chester, New York, NY 105783-4930. Claims for missing issues should be made immediately 
on receipt of the subsequent issues. 

Orders and enquiries for issues prior to 2002 should be sent to: Intercept Ltd., P.O. Box 716. Andover, 
Hampshire SP10 1YG, Telephone: (01264) 334748, Fax: (01264) 334058, Email: intercept@andover.co.uk, 
Internet: http://www.intercept.co.uk 

ADVERTISING Apply to the Publisher. Address enquiries to the Advertising Promoter. 

COPYRIGHT AND PERMISSIONS This journal is registered with the Copyright Clearance Center, 222 Rosewood Drive, 
Danvers, MA 01923, USA. Organizations in the USA who are also registered with CCC may therefore photocopy material 
(beyond the limits permitted by Section 107 and 108 of US Copyright law) subject to payment to CCC of the per-copy fee of 
$16.00. This consent does not extend to multiple copying for promotional or commercial purposes. Code 0968-0470/2002 
$16.00. ISI Tear Sheet Service, 3501 Market Street, Philadelphia, PA 19104, USA, is authorised to supply single photocopies 
of separate articles for private use only. Organizations authorised by the UK Copyright Licensing Agency may also copy 
material subject to the usual conditions. For all other use, permission should be sought from Cambridge University Press. 

No part of this publication may otherwise be reproduced, stored or distributed by any means without permission in writing 
from Cambridge University Press, acting for the copyright holder. 

Copyright © 2002 The Natural History Museum 

ELECTRONIC ACCESS This journal is included in the Cambridge Journals Online service which can be found at: 
http://journals.cambridge.org 

For further information on other Press titles access http://uk.cambridge.org or http://us.cambridge.org 

World list abbreviation: Bull. nat. Hist. Mus. Lond. (Zool.) 

ISSN 0968-0470 

The Natural History Museum Zoology Series 

Cromwell Road Vol. 68, No. 2, pp. 5 1-163 

London SW7 5BD Issued 28 November 2002 

Typeset by Ann Buchan (Typesetters), Middlesex 

Printed in Great Britain by Henry Ling Limited, at the Dorset Press, Dorchester, DTI 1HD 



Editorial: Garth Underwood - Dedication 




This issue, the last of the Zoology series of the Bulletin of the Natural 
History Museum, is dedicated to Dr Garth Underwood. Garth has had 
a long association with the Natural History Museum. In 1 964, he was 
appointed Principal Research Fellow to work on snake systematics, a 
project which culminated in the modestly titled "'A contribution to the 
classification of snakes" (Underwood, 1967). This book had a major 
impact on snake classification, pioneering the use of soft anatomy as 
a source of systematic characters. Its importance may be readily 
appreciated from the many references made to it in many of the papers 
in this special issue of the Bulletin (see especially Kochva in the 
introduction to his paper on burrowing asps, Atractaspis). In a more 
informal sense. Garth's association with the Natural History Museum 
started much earlier than 1964; a visit to the Museum in the late 1930's 
apparently gave him useful information for answering his Higher 
School Certificate papers in Zoology! Like many zoologists, an 
interest in natural history was something that was ingrained, and it 
seems that Garth always was seeking explanations for biological 
phenomena. His father, Leon Underwood, an eminent British sculp- 
tor and painter, dedicated a book called Animalia, subtitled Fibs 
about Beasts to Garth, showing him as a baby, thoughtfully looking at 
a frog. The book offers poetic or fanciful explanations about the 
animals within its pages, rather than scientific ones. The dedication 



reads: "To Garth, For whom cleaving facts asunder fall, And fancy 
sheds a healing light on all". Garth, if not then, certainly now seeks 
more objective, scientific interpretations in the biological sciences, 
particularly of snake relationships. 

Even a brief dedication such as this would be seriously deficient 
if it did not mention the contribution Garth has made to herpetology, 
not just in terms of his published work but through his encourage- 
ment and supervision of the studies of others. "A contribution to the 
classification of snakes" was a starting point; Garth has always 
sought new characters to shed new light on snake relationships, 
devised new ways of looking at data, and has never been afraid to 
revisit previous work to improve upon and revise earlier results. He 
has passed on these ideas to others; within the Museum alone he has 
supervised no less than 6 PhD's, most relating to snakes, but also 
encompassing frog and insect systematics. He has also run under- 
graduate and postgraduate courses in taxonomy, through times when 
systematics was less appreciated than formerly or even today. 

Many people owe Garth a considerable debt of gratitude for his 
help, guidance and support. He has been an inspiration to genera- 
tions of undergraduates, postgraduates and scientific colleagues 
worldwide; we hope he will be pleased with this token of our 
appreciation. 

Editors for this issue: 
Barry Clarke and Mark Wilkinson 



Sadly, Garth died on 15th October 2002 before this issue came out. He had seen or was aware of much of its contents. 

Photograph showing Garth Underwood in May 1966 when he was working on his "Contribution to the classification snakes". © The Natural History Museum 



X* C? 



Bull. nut. Hist. Mux. bond. (Zool.) 68(2): 51-55 



S(^^ 



ber2002 




71 



kistoft :um 
20 DEC 2002 



Hemipenial variation in the African snake 
genus Crotaphopeltis Fitzinger, 1843 
(Serpentes, Colubridae, Boiginae) 



THOMAS ZIEGLER 

Zoplogisches Forschungsinstitut und Museum Alexander Koenig, Adenauerallee 160, D-53U3 Bonn. 
Germany, e-mail: dr.th.ziegler@t-online.de 

JENS B0DTKER RASMUSSEN 

Zoological Museum. University of Copenhagen, Universitetsparken 15, DK-2 100 Copenhagen 0. Denmark, e- 



mail: jhrasmussen@zmuc.ku.dk 



SYNOPSIS. Hemipenial variation within the six recognized species of the African snake genus Crotaphopeltis is described. The 
hemipenial ornamentation of the widely distributed species. C. hotamboeia and C. hippocrepis, seems fairly constant whereas the 
ornamentation of C. degeni and C. tornieri, species with disjunct distributions, displays relatively large intraspecific variation. 
In spite of the variation, the size and ornamentation of the hemipenes serve to distinguish between sympatic and parapatric 
species. 



INTRODUCTION 



The various species within the African snake genus, Crotaphopeltis 
have been treated in a series of taxonomic papers (Rasmussen, 1 985: 
1993a; 1997: Rasmussen etal., 2000). Hemipenes have been exam- 
ined in situ and in everted or even in partly everted condition where 
possible. Recent collecting has increased the number of specimens 
with everted hemipenes. This provides a welcome opportunity for 
redescribing the hemipenes of the six recognized species. Further, 
differences between the hemipenes of the various isolated popul- 
ations of C. tornieri, of C. degeni, and of the various populations of 
the widely distributed C hotamboeia may provide an indication of 
the taxonomic status of some of the populations. 



MATERIAL AND METHODS 

Specimens included in the present study are held in the collections of 
the following museums: The Natural History Museum, London 
(BMNH); California Academy of Sciences, San Francisco (CAS); 
Museum of Comparative Zoology, Harvard (MCZ); Musee Royal 
d'AfriqueCentrale.Tervuren (MRAC);Rijksmuseum van Naturlijke 
Historic Leiden (RMNH); National Museum of Natural History, 
Washington (USNM); Zoologisches Forschungsinstitut und Mu- 
seum Alexander Koenig, Bonn (ZFMK); Zoological Museum, 
University of Copenhagen, Copenhagen (ZMUC). 

In the following list the catalogue numbers and places of origin are 
given for each specimen of the six currently recognized Crotapho- 
peltis species: 



Crotaphopeltis barotseensis. BOTSWANA: 
(ZMUC 631232). 



Okawanga Delta 



Crotaphopeltis braestrupi. SOMALIA: Mareri, ca. 8 km SW of 
Gelib(CAS 153370, 153379). 

Crotaphopeltis degeni. CAMEROON: Gueme (MRAC 73-15- 
R209). ETHIOPIA: Gambela (USNM 24389). SUDAN: Assalaya, 
10 mis E. of Kosti (RMNH 24411, 25018). TANZANIA: Tumba, 



Lake Rukwa (ZMUC 631233); Minziro Forest (ZMUC R631598, 
R631610, R631618. R631621). UGANDA: Kome Island, Victoria 
Nyanza(BMNH 1984.883). 

Crotaphopeltis hippocrepis. GHANA: Legon (ZFMK 63880); Legon 
Hill (ZMUC R631238); Legon Road, Achimota (ZFMK 63775); 
Wa, Secondary School (ZFMK 63875. 63877). 

Crotaphopeltis hotamboeia. CONGO: Tchissanga (ZMUC 
R631 177). GHANA: Wa (ZFMK 63874). KENYA: Langata, NW 
of" Nairobi (ZMUC R63984). SOUTH AFRICAN REPUBLIC: 
Cape Peninsula (ZMUC R63894); Tshaneni (ZMUC R63889). SU- 
DAN: Talanga Forest (ZMUC R63980); Torit (ZMUC R63979). 
TANZANIA: Magombero Forest (ZMUC R63921); Msolwa area, 
Rubeho Mountains (ZMUC R63 1205-06, R63 1208-10); Rungwe 
Mountains (ZMUC R631264, R631266, and R631268). ZAIRE: 6 
km NE of Kafumba, near Kikwit (ZMUC R63 1072). 

Crotaphopeltis tornieri. TANZANIA: East Usambara Mountains: 
Amani (R631 122-3); Kwamkoro (R631I27). Rungwe Mountains: 
Rungwe Mission ZMUC R631257. Udzungwa Mountains: Kihanga 
River, Udzungwa Scarp Forest Reserve ZMUC R63 1269-70; 
Kilanzi-Kitungulu Forest Reserve ZMUC R63 1244-45, R631252- 
54. R631256. West Usambara Mountains: Mazumbai (ZMUC 
R63963, R631129. R631135, R631137, R631140, R63 1142-3, 
R63 1146-7. R631155). 

Terminology follows Bohme (1988) and Ziegler & Bohme (1997). 
Preparation of the hemipenes of specimens previously preserved in 
alcohol was done according to the method described by Pesantes 
(1994) and Ziegler & Bohme (1997). 



RESULTS 

Crotaphopeltis barotseensis (Fig. 1). In situ hemipenes extend to 
subcaudal scute no. 7-8 (x=7.4, n=5) (Rasmussen, 1997). The only 
known everted hemipenes of this species have been prepared from a 
preserved specimen (ZMUC R631232). Consequently the organs 
are somewhat wrinkled, hardened and not completely distended. 



© The Natural History Museum, 2002 



52 



T. Z1EGLER AND J.B. RASMUSSEN 




Fig. 1 Crotaphopeltis barotseensis, right hemipenis in sulcal, left 
hemipenis in asulcal view of ZMUC R631232. 

Pedicel covered with tiny spines. Lower truncus covered with 
several somewhat enlarged spines, the most conspicuous on each 
side of the sulcus. Spines decrease in size towards the terminal 
somewhat calyculate apex. Sulcus spermaticus unforked, leading 
directly to the apex. 

Crotaphopeltis braestrupi (Fig. 2). In situ hemipenes extend to 
subcaudal scute no. 15-23 (x=18.4, n=61) and may be twice as long 
as in sympatric C. hotamboeia (Rasmussen, 1985). This distinctive 
difference is not reflected in the everted organs; however, the 
hemipenes of the present specimens (CAS 153370, 153379) are not 
entirely everted. 

Pedicel covered with tiny spines except for a longitudinal depres- 
sion on the outer asulcate surface that extends from the base of the 
pedicel to the lower truncus. Lower truncus covered with two to 
three slightly enlarged spines, one on either side of the sulcus and a 
less conspicuous one on the asulcate surface. Spines decrease in size 
towards the terminal somewhat calyculate apex. Sulcus spermaticus 
unforked, leading directly to the apex. 

Crotaphopeltis degeni (Fig. 3). In situ hemipenes extend to subcaudal 
scute no. 7-11 (x=8.4, n=35) (Rasmussen, 1997; Rasmussen et ai, 
2000). 




Fig. 2 Crotaphopeltis braestrupi, asulcal view of right hemipenis of 
CAS 153379. 




Fig. 3 Crotaphopeltis degeni, asulcal view of hemipenes of ZMUC 
R631621. 



HEMIPENIAL VARIATION IN CROTAPHOPELTIS 



53 



In the specimens from Minziro Forest, Tanzania, with freshly but 
not completely everted hemipenes (ZMUC R631598, R631610, 
R6316I8, R631621) superficial genital morphology is as follows: 
Pedicel covered with tiny spines. Lower truncus somewhat con- 
stricted and with two distinctly enlarged spines: one outside the 
sulcus spermaticus and another on the inner truncal surface. Re- 
maining spines slightly decreasing in size towards the apex. Sulcus 
unforked, running directly to the apex. 

In the specimen from Lake Rukwa, Tanzania (ZMUC R631233) 
the right hemipenis was everted after fixation. The somewhat wrin- 
kled, hardened and incompletely distended organ bears scarcely 
detectable, slightly enlarged spines on the lower truncus. In this 
specimen the sulcus is unforked, running directly to the apex, ending 
in a terminal extension. Tip of apex calyculate. 

In contrast, in the incompletely everted hemipenes of the two speci- 
mens of Crotaphopeltis degeni from Sudan (RMNH 2441 1, 25018) 
the lower truncus bears a ring of several distinctly enlarged spines. 

In the uneverted hemipenes of specimens from Ethiopia (USNM 
24389), Uganda (BMNH 1984.883) and Cameroon (RGMC 73-15- 
R209) clearly enlarged lower truncal spines are discernible. 

Crotaphopeltis hippocrepis (Fig. 4). In situ the hemipenes extend to 
subcaudal scute no. 8-12 (x= 10.1, n = 38) (Rasmussen etal, 2000). 
Pedicel covered with tiny spines except for a longitudinal depres- 
sion on the asulcate surface. Lower truncus with two distinctly 



fe, 






Fig. 5 Crotaphopeltis hotamboeia, asulcal view of hemipenes of ZMUC 
R631I77. 




Fig. 4 Crotaphopeltis hippocrepis, sulcal view of left hemipenis of 
ZFMK 63877. 



Fig. 6 Crotaphopeltis hotamboeia, asulcal view of hemipenes of ZMUC 
R631210. 

enlarged spines on either side of the sulcus (ZFMK 63875, 63877. 
ZMUC R631238), each followed above by several (usually 1-3) 
enlarged spines, apically decreasing in size (see also Rasmussen et 
al., 2000: fig. 3). Even in the only basally everted hemipenes of 
ZFMK 63775 and ZFMK 63880 these enlarged spines are easily 
recognizable. The remaining spines of truncus and apex are me- 
dium-sized, decreasing in size towards the apex which is calyculated 
terminally. Unforked sulcus spermaticus leading directly towards 
the apex, ending in a terminal extension. 

Crotaphopeltis hotamboeia (Figs 5,6). In situ hemipenes extend to 
subcaudal scute no. 7-14 (x=10.1, n=308). 

Pedicel of the hemipenis of C. hotamboeia covered with tiny spines 
except for a longitudinal depression on the asulcate surface. Lower 
truncus with three distinctly enlarged spines, one on each side of the 
sulcus, the third on the asulcate surface (ZFMK 63874, ZMUC 
R63889, R63979-80, R63 1 1 77, R63 1 264. 63 1 266, R63 1 268). Even 
in the only basally everted hemipenes of ZMUC R63894, R63984, 
and R631072 these three enlarged spines are also easily detectable. 
Distal to the three enlarged spines the hemipenis is covered with 
medium-sized spines decreasing in size towards the apex. The very tip 
of the apex seems to be somewhat calyculate. Unforked sulcus is 
leading directly to the apex ending in a terminal extension. 



54 



T. ZIEGLER AND J.B. RASMUSSEN 




Fig. 7 Crotaphopeltis tornieri, asulcal view of hemipenes of ZMUC 
R631147. 

Contrary to this condition the hemipenes of ZMUC R63 1205-6 
and R631208-10 (Msolwa area, Rubeho Mts., Tanzania) have, in 
addition to the three enlarged spines, a ring consisting of variously 
enlarged spines on the asulcate surface of the middle truncus. A 
similar condition was found in the hemipenes of ZMUC R63921 
(Magombero Forest, ca. 50 km S Mikumi, Tanzania). 

Crotaphopeltis tornieri (Fig. 7). In situ hemipenes extend to 
subcaudal scute no. 7-11 (x=8.4, n=27)(Rasmussen, 1993a; unpubl.). 

The pedicel of the hemipenis of C. tornieri is covered with tiny 
spines except for a longitudinal depression on the asulcate surface. 
Lower truncus usually with some enlarged spines. Distal to the 
enlarged spines, the spines become smaller towards the apex, which 
is terminally calyculate. Sulcus spermaticus is not forked and leads 
directly to the apex ending in a terminal extension. 

In specimens from East and West Usambara Mountains the largest 
spine is on the outside of the lower truncus, the second largest on the 
asulcate surface. The spine ornamentation of the hemipenes of the 
West Usambara population was somewhat variable, in the hemipenes 
of ZMUC R63963 (Rasmussen, 1993: fig. 2) no distinctly enlarged 
spine could be found on the asulcate surface. 

In some specimens (ZMUC R63 1 245 and R63 1 252) from Kilanzi- 
Kitungulu Forest Reserve, Udzungwa Mountains, the enlarged spine 
on the outside of the lower truncus is relatively small; furthermore 
an enlarged spine on the asulcate surface is scarcely detectable. In 
other specimens (ZMUC R63 1244, R63 1253-4 and R63 1256) from 
the same area, various enlarged spines are present on the lower 
truncus. 

In specimens from Kihanga River, Udzungwa Scarp Forest Re- 
serve (ZMUC 631269-70) and from Rungwe Mission, Mount 
Rungwe (ZMUC 631257), only the enlarged spine on the outside of 



the lower truncus is easily seen. On the asulcate surface some 
enlarged spines are present. 



DISCUSSION 

The hemipenial structures of Crotaphopeltis share several characters. 
e.g., pedicel largely covered with tiny spines, lower truncus with 
enlarged spines, spines decrease in size distally, apex calyculate, 
sulcus spermaticus unforked and leading directly to the apex. Regard- 
ing the simple and stout to elongate hemipenes within Crotaphopeltis, 
principally the differences in ornamentation of the (lower) truncus 
seem to offer a clue as to separate these taxa, in spite of the relatively 
large interspecific variation displayed. 

Crotaphopeltis barotseensis principally has one moderate en- 
larged spine on each side of the sulcus in the hemipenes of the single 
specimen examined. The species has a restricted distribution in 
Central Southern Africa and is easily distinguished from sympatric 
C. hotamboeia by general morphology (Rasmussen, 1985) as well 
as hemipenial morphology. 

Hemipenes of Crotaphopeltis braestrupi have two to three slightly 
enlarged spines on the truncus, one on each side of the sulcus and an 
inconspicuous one on the asulcate side. Comparing the hemipenes 
of C. braestrupi with those of sympatric C. hotamboeia from Kenya 
and Somalia, Rasmussen (1985) stated: 'The ornamentation is also 
very different. In C. braestrupi the hemipenis is covered with 
slender spines, of which three are somewhat enlarged. In C. 
hotamboeia the hemipenis is covered with stout spines of which 
three basal ones are strongly enlarged (Rasmussen, 1985: fig. 9).' It 
is difficult to judge the form of the variable spines (slender versus 
stout), which depend on the condition and the method of preparation 
of the hemipenes, but the three distinctly enlarged spines on the 
lower truncus are nonetheless characteristic of the hemipenes of C. 
hotamboeia across its entire distribution, i.e., Sub-Saharan Africa. 
Further studies are needed to show whether a separate taxon is 
justified for the specimens from Msolwa and Magombero Forest in 
Tanzania which, in addition to the three enlarged spines, have a ring 
consisting of some distinctly enlarged spines on the asulcate side of 
the hemipenes. External morphological investigations so far do not 
support such an assumption (Rasmussen, in prep.). 

The hemipenes of Crotaphopeltis degeni usually bear two (speci- 
mens from Tanzania) to several enlarged truncal spines (specimens 
from Cameroon, Ethiopia, Sudan, Uganda). Concerning the truncal 
spines of the hemipenes of Crotaphopeltis degeni from Sudan 
Rasmussen (1997) stated, 'up to six enlarged, stout spines, one each 
side of the sulcus and two to four (usually three) more or less 
enlarged spines on the asulcate aspect of the organ'. The lower 
truncal spines of the hemipenis of a specimen from Lake Rukwa, 
Tanzania appear only slightly enlarged, most probably due to ever- 
sion after fixation. Rasmussen (1997) observed only two enlarged, 
proximal spines in the incompletely everted hemipenes of speci- 
mens from Kenya and Uganda. Crotaphopeltis degeni apparently 
has a disjunct distribution like that of Causus resimus (Spawls and 
Branch, 1995). Despite the intraspecific variation of this taxon it is 
easily distinguished from sympatric C. hotamboeia and parapatric 
C. hippocrepis by hemipenial morphology and as well as general 
morphology (Rasmussen et al., 2000). 

The hemipenes of Crotaphopeltis hippocrepis are characterized 
by the possession of two distinctly enlarged spines, each followed 
by a row of accessory spines decreasing in size apically. This unique 
ornamentation of the hemipenes seems fairly constant within the 
entire distribution area (West Africa) of C. hippocrepis and the 
species seems to be homogeneous (Rasmussen et ah, 2000). 



HEMIPENIAL VARIATION IN CROTAPHOPELT1S 



55 



In the northern part of its distribution (The Usambara Mountains) 
the hemipenes of Crotaphopeltis tornieri usually have two slightly 
enlarged spines, one on the outside of the truncus and one (rarely 
missing) on the asulcate side. Based on external morphology, 
Rasmussen (1993) found significant differences between the popul- 
ations of C. tornieri from East and West Usambara Mountains. The 
present results of the genital investigation do not lend support to the 
recognition of different forms, nor do molecular (Gravlund. 2002) or 
microdermatoglyphic (Berggreen, 1996) studies. Accordingly, the 
differences found in numbers of ventral and caudal scutes between 
these areas are probably due to genetic drift in small, isolated 
populations. 

Gravlund (2002) found significant differences between the mole- 
cular composition of the northern and southern populations of C. 
tornieri in the Eastern Arc. In particular, the population from the 
Rungwe Mountains is very different molecularly from those of the 
Udzungwa and the Usambara Mountains. The present study also 
shows slight hemipenial differences between these populations. In 
the hemipenes of the populations from Kihanga River (Udzungwa 
Scarp Forest Reserve) and the Rungwe Mountains only the enlarged 
spine on the outside of the truncus is visible. The picture, however, 
becomes blurred as hemipenes of specimens from Kilanzi-Kitungulu 
(Udzungwa Mountains) may have either a relatively small spine on 
the outside and likewise on the asulcate side or various enlarged 
spines on the truncus. Thus, the data on genital morphology indicate 
that the spine ornamentation of the lower truncus in the hemipenes 
of C. tornieri is variable even within single populations and with 
current knowledge should not be used for defining different taxa 
within the C. tornieri complex. This is in accordance with Berggreen 
(1996) who found some microdermatoglyphic variation within and 
between the various populations. The variation, however, was not 
population specific. Crotaphopeltis tornieri (s.l.) is restricted to the 
montane forest of the Eastern Arc of Tanzania and is easily distin- 
guished from sympatric/parapatric C. hotamboeia from the lowland 
savanna. 

Thus, irrespective of the fact that intraspecific variation may 
occur within the various species of Crotaphopeltis, the ornamenta- 
tion of the lower truncus serves at least to distinguish between the 
males of sympatric or parapatric species. The length at least of the 
inverted hemipenes may also serve to distinguish between the 
species and so may the unique rows of accessory spines in the 
hemipenes of C hippocrepis. Here, it is interesting to note, that the 



pattern of the latter species is similar (Dowling, in litt.) to that of the 
sister-genus Dipsadoboa (Rasmussen, 1979), which is also charac- 
terized by similar, highly derived genital morphological features 
(Rasmussen, 1993b). 



ACKNOWLEDGEMENTS. For the loan and making available material for 
examination we thank C. McCarthy (BMNH), R. Drewes (CAS), J. Rosado 
(MCZ), D. Meirte (MRAC), M. Hoogmoed (RMNH). L. Ford (USNM), and 
W. Bohme (ZFMK). We are grateful to B.T. Clarke and two anonymous 
referees for commenting on the manuscript. Support for travel for the junior 
author to Tervuren, Bonn, and London has been gratefully received from the 
Danish National Research Council, Grant no. 56043. 



REFERENCES 



Berggreen, A. 1996. Microdermatoglyphics of snake scales. Unpublished Master's 

thesis. University of Copenhagen. 78p + 2 appendices. 
Bohme, W. 1988. Zur Genitalmorphologie der Sauria: funktionelle und 

stammesgeschichtliche Aspekte. Bonner zoologische Monographien 27: 1-176. 
Gravlund, P. 2002. Molecular phylogeny of Tornier's cat snake (Crotaphopelris 

tornieri), endemic to East African mountain forests: biogeography. vicariance events 

and problematic species boundaries. Journal of Zoological Systematics and Evolu- 
tionary Research 40: 46-56. 
Pesantes, O.S. 1994. A method for preparing the hemipenis of preserved snakes. 

Journal of Herpetology 28: 93-95. 
Rasmussen, J.B. 1979. An intergeneric analysis of some boigine snakes - Bogert's 

( 1940) Group XIII and XIV (Boiginae, Serpentes). Videnskabelige Meddelelser fra 

dansk Naturhistorisk Forening 141: 97-155. 
1985. A new species of Crotaphopeltis from East Africa, with remarks on the 

identity of Dipsas hippocrepis Reinhardt, 1843 (Serpentes: Boiginae). Steenstrupia 

11: 113-129. 
1993a. The current taxonomic status of Tornier's cat-snake (Crotaphopeltis 

tornieri). Amphihia-Reptdiu 14: 395-409. 
1993b. A taxonomic review of the Dipsadoboa unicolor complex, including a 

phylogenetic analysis of the genus (Serpentes. Dipsadidae, Boiginae). Steenstrupia 

19: 129-196. 
1997. On two little known African water snakes [Crotaphopeltis degeni and C. 

baroiseensis). Amphibia-Reptilia 18: 191-206. 
, Chirio, L., & Ineich, I. 2000. The Herald Snakes {Crotaphopeltis) of the Central 

African Republic, including a systematic review of C. hippocrepis. Zoosystema 22: 

585-600. 
Spawls, S. & Branch, B. 1995. The dangerous snakes of Africa. Natural History: 

Species Directory: Venoms and Snakebite. 192p. Blandford. London. 
Ziegler, T. & Bohme, W. 1997. Genitalstrukturen und Paarungsbiologie bei squamaten 

Reptilien, speziell den Platynota. mit Bemerkungen zur Systematik. Mertensiella. 

Rheinbach 8: 1-207. 



*'X f3S6^S^. 



Bull. not. Hist. Mas. bond. (Zool.) 68(2): 57-74 



Issued 28 November 2002 



Review of the Dispholidini, with the 
description of a new genus and species from 
Tanzania (Serpentes, Colubridae) 

D.G. BROADLEY 

Research Associate, Natural History Museum of Zimbabwe, Bulawayo, Zimbabwe 

Present address: Biodiversity Foundation for Africa, P.O. Box FM 730 Famona, Bulawayo, Zimbabwe, email: 

broadley @ telconet. co.zw 

V. WALLACH 

Museum of Comparative Zoology, Harvard University, 26 Oxford Street, Cambridge, MA 02138, U.S.A. email: 
vwallach@oeb. harvard, edu 

SYNOPSIS. The tribe Dispholidini (Bourgeois 1968) is reviewed, paying particular attention to dentition and visceral anatomy. 
A new genus and species, Xyelodontophis uluguruensis, is described from the Uluguru Mountains in Tanzania. All five genera 
have enlarged rear maxillary teeth. Thrasops seems to be basal, Rhamnophis shows the development of dagger-like teeth tapering 
from base to tip. then the other three genera appear to radiate, with Xyelodontophis having more derived dagger teeth broadest 
in the middle, while Dispholidus and Thelotornis seem to have independently developed enlarged grooved rear fangs. 

Thrasops schmidti is recognised as a relict evolutionary species. No subspecies of Rhamnophis aethiopissa or Dispholidus 
typus are recognised, but the population of Dispholidus on Pemba Island probably represents an undescribed species. 



INTRODUCTION 



MATERIALS AND METHODS 



From the time of Boulenger's catalogues ( 1 893-96), it was custom- 
ary to separate the aglyphous colubrid snakes (subfamily Colubrinae) 
from the opisthoglyphs (subfamily Dipsadomorphinae or Boiginae). 
see for example Loveridge (1957) and FitzSimons (1962). 

Bourgeois (1968) erected a subfamily Dispholidinae, including 
the aglyphous genera Thrasops Hallowell 1857 and Rhamnophis 
Giinther 1862, and the opisthoglyphous genera Dispholidus 
Duvernoy 1 832 and Thelotornis A. Smith 1 849. Subsequent authors 
have often treated Rhamnophis as a synonym of Thrasops (e.g. 
Hughes & Barry, 1969; Pitman, 1974; Spawls, 1978; Hughes, 1983; 
Trape & Roux-Esteve, 1995; Chippaux, 1999) and many have 
considered these aglyphous snakes to be members of the tribe 
Philothamnini (e.g. Dowling & Duellman, 1978). 

During a review of the genus Thelotornis in East Africa (Broadley, 
2001), a snake from montane forest on the summit of the Uluguru 
Mountains was initially assumed to represent a new species. How- 
ever, examination of its rear maxillary teeth showed that they were not 
the anticipated grooved fangs, but distinctive curved dagger-shaped 
teeth with sharp anterior and posterior ridges, which are widest 
midway along the tooth. To determine the relationships of this strange 
snake to the other taxa of the Dispholidini, its skull was prepared (after 
examination of the dental gland) and compared with skulls of the other 
genera. This prompted a review of the genera Thrasops and 
Rhamnophis, which appear to represent basal taxa of the Dispholidini. 
As the 'Dagger-tooth Vine Snake' of the Uluguru Mountains seems to 
be transitional between Rhamnophis and Thelotornis, but cannot be 
accommodated in any of the existing genera of the tribe Dispholidini, 
it is proposed to erect a new genus and species for it. In external 
appearance and scale counts it resembles Thelotornis, but it lacks the 
distinctive horizontal key-hole shaped pupil of that genus. 

It is with pleasure that we dedicate this paper to Garth Underwood, 
in recognition of the major contributions that he has made to our 
understanding of African snakes. 



This study is largely based on material available in the Natural 
History Museum of Zimbabwe and the Museum of Comparative 
Zoology, Harvard, with additional data derived from the literature. 
Unfortunately the collections in the Natural History Museum were 
not accessible. Loveridge's 1944 revisions of Thrasops, Rhamnophis 
and Thelotornis were based largely on scalation, supplemented by 
maxillary tooth counts and coloration of head and neck in the case of 
Thelotornis. We have emphasised the morphology of the rear max- 
illary teeth and skull, and have also used data from the visceral 
anatomy, using the Philothamnini as the outgroup for comparative 
purposes. Data for good series of Philothamnus angolensis, 
Hapsidophrys lineatus and H. smaragdina were available [Broadley 
(1966) provisionally synonymised Gastropyxis Cope 1860 with 
Hapsidophrys Fischer 1856, and this move is supported by the 
visceral data]. In the species accounts we have only presented 
chresonymies, full synonymies are included in the review of the East 
African Thelotornis (Broadley, 2001) and investigation of the vari- 
ation in the wide-ranging genus Dispholidus awaits future workers! 

In the description of the visceral anatomy, the mean value for most 
characters as % snout-vent length (SVL) is presented first, followed 
parenthetically by the range or midpoint (MP) value. When only the 
name of an organ is given, the value represents its length. Ratios of 
two visceral characters are presented in fractional notation. When 
only one value is given for a character, it is identical in the two 
specimens or differs by less than 0.1%. 

The position of the umbilicus is determined by the most anterior 
ventral bearing a scar (the scar usually covers three ventrals and the 
umbilicus exited through the medial scute). The umbilical scar-vent 
interval is calculated by dividing the number of ventrals from the 
scar to the vent by the total number of ventral scutes. 

Material for which skulls or viscera were examined is listed in 
appendices. Institutional abbreviations follow Leviton etal. (1985), 
with the addition of: 



© The Natural History Museum, 2002 



58 



D.G. BROADLEY AND V. WALLACH 



IRSL = Instituit d'Rechereche Scientifique, Lwiro, Democratic 
Republic of Congo (DRC); UNAZA = Universite National du Zaire, 
Kisangani, DRC; VW = Van Wallach dissection number (museum 
deposition of specimen unknown). 



CHARACTER ANALYSIS 

1. Rear maxillary teeth. The three rear maxillary teeth of Thrasops 
flavigularis (type species) and T. jacksonii are enlarged and sepa- 
rated from the small anterior teeth by a diastema, they taper from 
base to tip and have slight ridges anteriorly and posteriorly (Fig. 1 A, 
Group B dentition of Jackson & Fritts, 1995). The posterior ridge 
becomes blade-like in some genera, e.g. Heterodon (Kardong, 1979), 
Thamnophis (Wright, etal, 1979) mdStegonotus (Jackson & Fritts, 
1995). 

The same teeth in Rhamnophis aethiopissa (Fig. IB) and R. batesi 
(Fig. 1 C) are curved, with sharp anterior and posterior ridges, but not 
nearly as well developed as in the 'Dagger-tooth Vine Snake' of the 
Uluguru Mountains, in which the ridges are broadest midway along 
the tooth, which is leaf-shaped, narrowing at the base (Fig. IF). In 
Dispholidus and Thelotornis the three greatly enlarged rear maxil- 
lary teeth are deeply grooved (Group D dentition of Jackson & 
Fritts, 1995), but these genera retain a strong ridge on the anterior 
face of the fangs. In Thelotornis, this ridge arises within the groove, 
so that the venom canal is divided, before petering out well before 
the fang tip (Fig. 1G, after Meier, 198 1 , fig 4). In Dispholidus on the 
other hand, the ridge runs along the anterior edge of the groove (Fig. 
1H, after Meier, 1981, fig 2). 

With regard to number of maxillary teeth, Rhamnophis aethiopissa 
(16 to 20 + 3) resembles Thrasops spp. (17 to 18 + 3), but R. batesi 
(30 to 35 + 3, Fig. 1C) is divergent in this respect. The Dagger-tooth 
Vine Snake has 14 + 3, thus matching Thelotornis (11 to 16 + III) in 
actual tooth number. Dispholidus shows a marked reduction in 
number of anterior teeth to 4 to 8 + III. Counts of maxillary tooth 
sockets are much higher in the Philothamnini: 1 7^48 in Philothamnus 
and 20-33 in Hapsidophrys. 

2. Dental (Duvernoys ') gland. This gland is small in Thrasops and 
Rhamnophis (Kochva, 1978), larger in Thelotornis (but smaller than 
the orbit), still larger in the Dagger-tooth Vine Snake (subequal to 
the orbit) and reaches its maximum development in Dispholidus 
(Kochva, 1978), with a large, purely serous, slightly branched, 
tubuloacinous Duvernoy's gland, the tubule walls highly folded, 
increasing the storage space within the gland (Taub, 1967). thus 
constituting a reservoir (Underwood, 1997). The mechanism of 
delivery of toxic dental gland secretions by low pressure systems 
has been demonstrated for Boiga irregularis (Kardong & Lavin- 
Murcio, 1993) and is effective regardless whether or not the teeth are 
grooved (Weinstein & Kardong, 1994). 

3. Skull. The Dispholidini were first recognised (as a subfamily) 
by Bourgeois (1968) on the basis of their similar skull morphology 
(Fig. 2). She drew attention to the forked ectopterygoid, large optic 
fenestra and interorbital vacuity (also noted by Underwood, 1967). 
The ectopterygoid is shallowly forked in Thrasops, Thelotornis and 
the Dagger-tooth Vine Snake, but is very deeply forked in both 
species of Rhamnophis and in Dispholidus. Underwood (1967) 
noted the absence of a Vidian canal in the skulls of Thrasops and 
Thelotornis, but Vaeth (1982) found a short, but distinct, Vidian 
canal in the skulls of three Thrasops jacksonii. 

4. Pupil shape. The pupil is round in Thrasops and Rhamnophis 
(Fig. 3) as in the Philothamnini, but in Dispholidus and the Dagger- 



tooth Vine Snake it may be more pear-shaped, due to an anterior 
prolongation. Thelotornis is distinguished by its horizontal 'key- 
hole' shaped pupil (Fig. 4B-D). 

5. Visceral anatomy. The Dispholidini can be characterized by the 
following visceral characteristics (Tables 2-\): umbilical scar-vent 
interval 8-12% total ventrals; hyoid short with posterior tips at 7- 
10% SVL, heart short, 1.5-3.1% (mean 2.4%); right systemic arch 
reduced to 0.20-0.40 left systemic arch diameter; liver narrow with 
midpoint at 43^t6% SVL; gall bladder craniad of pancreas and 
spleen; testes normally unipartite but occasional specimens with bi- 
or tri-partite organs (the additional segments being small sections 
separated from the main body either posteriorly or anteriorly); 
kidneys compact but segmented (15—45 segments); no tracheal 
lung; trachea with narrow, well-separated cartilages that lack free 
tips, tracheal membrane expanded to 2.0-4.0 (mean = 2.9) times the 
circumferential width of the rings; weak development of the cardiac 
lung to midheart level; tracheal entry into right lung subterminal, 
right lung with small anterior lobe and small orifice; right lung 
elongate (69-70% SVL), extending to 94-97% body length, cranial 
vascular portion 0.15-0.25 lung length, usually with midventral 
avascular strip, caudal saccular portion long 0.75-0.85 lung length; 
faveolar parenchyma arranged in 2-3 tiers with pattern of transverse 
smooth muscle ribs enclosing rows of paired faveoli; semisaccular 
portion of lung short (0.10-0.20 vascular lung length) with abrupt 
termination of parenchyma along a transverse border. 

The hemipenes of the genera Thrasops, Dispholidus and 
Thelotornis appear to be similar, being simple, capitate, with an 
undivided sulcus. There are large basal spines which diminish in size 
distally and are replaced by calyces on the distal cap (Bogert, 1940). 
The organs of the Dagger-tooth Vine Snake show little difference, 
the nude basal portion has four large hooks, the medial portion is 
spinose and the apical portion is calyculate. The hemipenes of 
Rhamnophis have not yet been described. 

6. Dorsal head coloration. The development of complex head 
patterns may aid in species recognition. All four species of Thrasops 
have the head uniform olive when subadult, eventually becoming 
uniform black. Rhamnophis batesii has a uniform brown or black 
head, while that of/?, aethiopissa is green, with the shields margined 
with black. A somewhat similar black vermiculation on a yellow or 
green ground is found in males of some populations of Dispholidus 
typus, but many have no colour pattern. The Dagger-tooth Vine 
Snake has dark margins to the head shields and yellow labials. The 
four species of Thelotornis can be distinguished by the colour 
pattern of the head (Broadley, 2001 ). The top of the head is uniform 
green in T kirtlandii (Fig. 4B), T usambaricus and some T. 
mossambicanus, but blue-green with black and pink speckling in T 
capensis (Fig. 4D). The temporals are uniform green in T kirtlandii 
(Fig. 4B) and T usambaricus, brown with black speckling in T 
mossambicanus (Fig. 4C), and pink margined with black in T 
capensis (Fig. 4D). The supralabials are uniform or with faint green 
or grey stippling in T kirtlandii, but the other taxa have black spots, 
usually including a speckled black triangle on the sixth labial. 

7. Throat pattern. All members of the Dispholidini (and some 
members of the Philothamnini) can inflate the throat in a threat 
display, reaching its maximum development in Dispholidus. 
Chippaux (1999, PI. 17) illustrates this phenomenon in Thrasops 
flavigularis, where the black dorsum contrasts with the pale throat, 
but in T. jacksonii the throat often becomes entirely black. 
Rhamnophis has the dark green dorsal scales bordered with black, 
the throat is yellowish in R. batesii and green in R. aethiopissa. 
Dispholidus comes in a wide range of colour patterns, but usually 



REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



59 





v-yzOfy 



-^r-^ryr)) 



<^ 




-^^J K^w^ 




^SJ^gas^^^^ 



® 






Fig. 1 Dentition: Left - maxillae of: A. Thrasops flavigularis; B. Rhamnophis aethiopissa; C. Rhamnophis batesii; D. Thelotomis kirtlandii; E. 

Dispholidus typus. Right - teeth of: F. Xyelodontophis uluguruensis; G. Dispholidus typus; H. Thelotomis kirtlandii. (A, C, D, E after Chippaux, 1999; B 
after Bourgeois, 1968: F, G, H after Meir, 1981). 



60 



D.G. BROADLEY AND V. WALLACH 







© 




Fig. 2 Skulls of: A. Xyelodontophis uluguruensis; B. Thelotornis mossambicanus; C. Rhamnophis batesii; D. Dispholidus typus. 



has a black spot on the side of the neck (Broadley, 1983, fig. 144), in 
this species the inflation may extend half way down the body. The 
inflated throat of Thelotornis is grey-white with distinctive black 
markings - crossbars in T. kirtlandii, chevrons in T. usambaricus, 
one or two elongate blotches in T. mossambicanus and two larger 
dorsally extensive blotches in T. capensis. 

8. Temporals and occipitals. In Thrasops there are almost invari- 
ably 1 + 1 temporals and there are no enlarged occipital shields. In 



Rhamnophis there is a single large temporal: R. batesii has four large 
occipitals, while R. aethiopissa has two very large ones. In 
Dispholidus, Thelotornis and the Dagger-tooth Vine Snake there are 
usually 1 + 2 temporals and three occipitals (or two separated by a 
smaller interoccipital). The Philothamnini tend to have more numer- 
ous temporals (1 + 1 up to 2 + 2 + 2) and no enlarged occipitals. 

9. Supralabials (Table 1). Thrasops usually has 8 supralabials, the 
fourth and fifth entering the orbit. Rhamnophis batesii has 7 or 8, 4 



REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



61 













Fig. 3 Head shields of: A. Thrasops flavigularis; B. Thrasops occidentalis, with midbody scalation to the right, compared with midbody scalation of T. 
flavigularis on the far right (after Parker, 1940); C. Thrasops jacksonii; D. Rhamnophis aethiopissa; E. Rhamnophis batesii. 



62 



D.G. BROADLEY AND V. WALLACH 












Fig. 4 Head shields of: A. Xyelodontophis uluguruensis (holotype); B. Thelotornis kirtlandii; C. Thelotornis mossambicanus; D. Thelotomis capensis; E. 
Dispholidus typus. 



REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



63 



Table 1 Dispholidini compared with Philothamnini: variation in midbody scale rows, ventrals, subcaudals and supralabials (rare variations in 
parentheses). 



Taxon 


midbody rows 


ventrals 


subcaudals 


supralabials [in orbit] 


Philothamnus spp. * 


(13) 15 


135-213 


60-175 


8/9 [4 to 6] 


Hapsidophrys spp. ** 


15 


150-176 


90-172 


8/9 [4 to 6] 


Thrasops flavigularis 


13(15) 


195-215 


128-146 


8 (9) [4,5 (5,6)] 


Thrasops occidentalis 


15- 19(21) 


175-187 


119-140 


8 (7) [4,5 (5,6)] 


Thrasops jacksonii 


(17) 19(21) 


187-211 


130-155 


8 (9) [4,5 (5,6)] 


Thrasops schmidti 


(16)17(19) 


168-184 


121-149 


8 [4,5] 


Rhamnophis aethiopissa 


15- 17(19) 


154-179 


117-159 


6-9 [3.4; 4,5; 5,6] 


Rhamnophis balesii 


13 


163-179 


91-114 


7-8 [4.5: 5.6] 


Xyelodontophis uluguruensis 


19 


168-169 


? 


8 [4.5] 


Thelotornis kirtlandii 


19 


162-189 


139-161 


8 (7, 9) [4,5 (5.6)] 


Thelotornis usambaricus 


19 


156-169 


151-175 


8 (9) [4,5 (3,4.5)] 


Thelotornis mossambicanus 


(17) 19(23) 


149-166 


127-158 


8 [4,5] 


Thelotornis capensis 


19 


144-177 


128-165 


8 [4,5] 


Dispholidus typus 


(17) 19(21) 


164-201 


104-142+ * 


7 (3,4) 



*Data from Hughes (1985). **Data from Chippaux ( 1999). "168 in D. 'pemba' (MCZ 45587). 



& 5 or 5 & 6 entering orbit. The widespread R. aethiopissa is more 
variable, 6 to 9 labials, often with 3 & 4 entering orbit in southern 
and eastern populations. Thelotornis and the Dagger-tooth Vine 
Snake usually have 8 (4 & 5) and Dispholidus 7 (3 & 4). 

10. Dorsal scales (Table 1 ). In Thrasops the dorsals are smooth in 
juveniles, the median rows keeled in adults, number of rows at 
midbody varies from 13-15 in T. flavigularis (which has the dorsals 
twice as long as the ventrals) to 17-21 (usually 19) in T. jacksonii. In 
Rhamnophis the dorsals are smooth, with the vertebral row enlarged, 
13 rows in R. batesii and 15-19 rows in R. aethiopissa. The dorsals 
are feebly keeled and usually in 19 rows in Thelotornis and the 
Dagger-tooth Vine Snake, while Dispholidus usually has 19 rows of 
strongly keeled scales. In the Philothamnini there are usually 15 
scale rows, which are usually smooth in Philothamnus, but keeled in 
Hapsidophrys. 

11. Ventral counts (Table 1). The highest counts are found in 
Thrasops flavigularis and the lowest in Thelotornis capensis. 

12. Subcaudal counts (Table 1). The lowest counts are found in 
Rhamnophis batesii and some populations of Dispholidus typus, 
while the highest are found in the two forest species of Thelotornis. 



SYSTEMATIC ACCOUNT 

Thrasops flavigularis ( Hallowell) 

Yellow-throated Bold-eyed Tree Snake 

Dendrophis flavigularis Hallowell, 1852, Proc. Acad. not. Sci. 

Philadelphia: 205. Type locality: 'Liberia', later corrected to 

Gabon. 
Hapsidophrys niger Giinther, 1872, Ann. Mag. not. Hist. (4) 9: 25. 

Type locality: Gaboon. 
Thrasops pustulatus Buchholz & Peters, 1875, Monatsb. Akad. 

Wiss. Berlin: 199. Type locality: Mungo, Cameroon. 
ThrasopsflavigularisBocage, 1895: 97; Bogert, 1940: 58; Loveridge, 

1944: 132; Trape & Roux-Esteve, 1995: 40; Chippaux, 1999: 95. 
Thrasops flavigularis flavigularis Stucki-Stirn, 1979: 319. 
Thrasops flavigularis stirnensis Stucki-Stirn, 1979: 632. 

Diagnosis. Dorsal scales in 13-15 rows at midbody, the dorsals 
much longer than the ventrals; ventrals 191-214; subcaudals 128- 
146; usually 2 labials in contact with the lowest postocular; no 
enlarged occipitals. 



DESCRIPTION. Supralabials 8 (rarely 9), fourth & fifth (rarely fifth 
& sixth) entering orbit; infralabials 9-12, the first 3-5 in contact 
with anterior sublinguals; preoculars 1 or 2; postoculars 3 (rarely 2), 
usually 2 labials in contact with the lowest; temporals 1 + 1; no 
occipitals. Dorsals in 17-15-13, 17-13-11, 15-13-13, 15-13-11 or 
13-13-11 rows, feebly keeled in adults; ventrals 191-214; cloacal 
divided; subcaudals 128-146 pairs. 

Coloration in life. Subadults olive to dark brown above, head 
uniform, body mottled with black and yellow, the black being on the 
interstitial skin and bases of the scales, the yellow in the centres of 
the scales, the yellow spots very pronounced on the tail. Chin and 
throat yellow, rest of venter chequered black and yellow. Adults 
usually uniform black above, venter blackish, but throat usually 
yellow or brownish white. 

SIZE. Largest 6 (IFAN 687 - Sibiti, Congo-Brazzaville) 1514 + 
586 = 2100 mm (Villiers, 1966); largest 9 (AMNH 50573 - Metet, 
Cameroon) 1235 + 505 = 1740 mm (Bogert. 1940). Stucki-Stirn 
(1979) gives the maximum length as 240 cm. 

Habitat. Lowland forest. 

Distribution. Southwestern Nigeria, Bioko Island, Cameroon, 
Gabon, Congo-Brazzaville, extreme eastern Democratic Republic 
of Congo and northwestern Angola (Fig. 5). 

Thrasops occidentalis Parker 

Western Bold-eyed Tree Snake 

Thrasops occidentalis Parker. 1940, Ann. Mag. nat. Hist. (11) 5 
273, fig. 1 & 2a. Type locality: Axim, Gold Coast [= Ghana] 
Loveridge, 1944: 131; Cansdale, 1961: 31, PI. vi, fig. 11 & 12 
Hughes & Barry, 1969: 1018; Chippaux, 1999: 100. 

Diagnosis. Dorsal scales in 1 5-2 1 rows at midbody, the vertebral 
row widened; ventrals 175-187; subcaudals 119-140; 3 labials in 
contact with lowest postocular. 

DESCRIPTION. Supralabials 8 (rarely 7 or 9), the fourth & fifth 
(rarely fifth & sixth) entering orbit; infralabials 8-10, the first 4-6 in 
contact with anterior sublinguals; preocular 1; postoculars 3, 3 
labials in contact with the lowest; temporals 1 + 1; no occipitals. 
Dorsals in 15-21 rows at midbody, the median rows keeled in adults, 
smooth in juveniles; ventrals 175-187; cloacal divided; subcaudals 
119-140 pairs. 

COLORATION IN LIFE. Juveniles with head and neck olive, body 



64 



D.G. BROADLEY AND V. WALLACH 



chequered in black and yellow above and below. Adults black 
above, chin and throat pale yellow, rest of venter dark olive. 

Size. Largest 6 (BMNH 66.1.28.6 - Sierra Leone, paratype) 670 
+ 495 = 1165 mm; largest 9 (BMNH 1911.6.30.2 - Axim, Ghana, 
holotype) 682 + 403 = 1085 mm. Cansdale (1961) states that this 
species can exceed 210 cm. 

Habitat. Lowland forest. 

Distribution. Guinea east to southwestern Nigeria (Fig. 5). 

Thrasops jacksonii Giinther 

Jackson's Bold-eyed Tree Snake 

Thrasops Jacksonii Giinther, 1895, Ann. Mag. nat. Hist. (6) 15: 528. 
Type locality: Kavirondo, Kenya. 

Rhamnophis jacksonii Boulenger, 1896: 632. 

Thrasops Rothschildi Mocquard, 1905, Bull. Mus. natn. Hist. nat. 
11: 287. Type locality: 'Afrique orientale anglaise'. 

Thrasops jacksonii jacksonii Loveridge, 1936: 249, 1944: 134 & 
1957: 264; Bogert, 1940: 58; Witte, 1953: 200; Laurent, 1956: 
187, 354 & 1960: 46; Roux-Esteve, 1965: 66, fig. 17; Villiers, 
1966: 1739; Bourgeois, 1968: 124, 278, fig. 51; Pitman, 1974:99, 
PI. G, fig. 4; Spawls, 1978: 5; Broadley, 1991: 532; Hinkel, 1992: 
319, PI. 306; Trape & Roux-Esteve, 1995: 40. 

Diagnosis. Dorsal scales in 19 (very rarely 17 or 21) rows at 
midbody; ventrals 187-214; cloacal divided; subcaudals 129-155; 
usually two labials in contact with lowest postocular. 

Variation. Supralabials 8 (rarely 9, very rarely 7), fourth and 
fifth (rarely fifth and sixth) entering orbit; infralabials 9-13, the first 
4-6 in contact with anterior sublinguals; preoculars 1-2 (rarely 3); 
postoculars 3 (very rarely 2 or 4), usually 2 labials in contact with 
lowest; temporals 1 + 1 (very rarely 1 + 2); no occipitals. Dorsals 
keeled in 19 (very rarely 17 or 21) rows at midbody; ventrals 181- 
214; cloacal divided; subcaudals 129-155 pairs. 

Coloration in life. Subadults dark olive above, mottled with 
black and buff posteriorly, greenish yellow below, becoming cheq- 
uered black and yellow posteriorly. Adults uniform black above and 
below, or with the throat yellow or greyish. Iris of eye black. 

Size. Largest 6 (AMNH 12288) 1320 + 580= 1 900 mm. largest 5 
(AMNH 12290) 1550 + 610 = 2160 mm, both from the Ituri Forest, 
Orientale Province, D.R.C. (Schmidt, 1923). Pitman (1974) puts the 
maximum length at about 2300 mm. 

Habitat. Rain forest and gallery forests from about 200 m in the 
lower Congo region to 2400 m on Mount Elgon (Pitman, 1974). 

Distribution. From the lower Congo, east through the Congo 
basin to southern Central African Republic, southern Sudan, Uganda, 
western Kenya and northwestern Zambia (Broadley, 1991) (Fig. 5). 

Thrasops schmidti Loveridge 

Schmidt's Bold-eyed Tree Snake 

Thrasops jacksonii schmidti Loveridge, 1936, Proc. biol. Soc. Wash- 
ington 49: 63. Type locality: Meru Forest, Mount Kenya, Kenya; 
1944: 137 & 1957: 264; Spawls, 1978: 5. 

Diagnosis. Dorsal scales in 17 rows; ventrals 168-184; subcaudals 
121-149; two labials in contact with lowest postocular. 

DESCRIPTION. Supralabials 8, the fourth and fifth entering the 
orbit; infralabials 10-12, the first 4 or 5 in contact with anterior 
sublinguals; preocular 1; postoculars 3, the lowest in contact with 2 



labials; temporals 1 + 1; no occipitals. Dorsals in 17 (rarely 19) rows 
at midbody, faintly keeled; ventrals 172-184; cloacal divided; 
subcaudals 121-147 pairs. 

Coloration in life. Subadult olive brown above, greyish white 
below, subcaudals grey. Adults uniform black. 

Size. Largest 8 (MCZ 9276 - Meru Forest, Kenya, holotype) 700 
+ 365 = 1065 mm; largest? (NMK 1222 - Embu Forest, Kenya) 
1 200 + 455 = 1 655 mm; largest unsexed (formerly NMK - Muthaiga, 
Nairobi, Kenya, paratype) 1671 +584 = 2255 mm (Loveridge, 1923. 
1936). 

Habitat. Montane forest. 

Distribution. Forests of the Kenya highlands from Mount Kenya 
south to Nairobi (Fig. 5). 

Remarks. T schmidti is readily diagnosable on ventral counts 
and is separated from the population of T jacksonii in the Kakamega 
Forest by 300 km, including the dry rift valley, so it is considered to 
represent an independently evolving taxon. 



Rhamnophis aethiopissa Giinther 

Splendid Dagger-tooth Tree Snake 

Rhamnophis aethiopissa Giinther, 1862, Ann. Mag. nat. Hist. (3) 9: 

129, PI. x. Type locality: West Africa; Roux-Esteve, 1965: 65, fig. 

16;Chippaux, 1999:97. 
Thrasops splendens Andersson, 1901, Bihang Till K. Svenska Vet.- 

Akad. Handl. 27(5): 11, PI. 1, fig. 8. Type localities: Bibundi & 

Mapanja, Cameroon. 
Rhamnophis ituriensis Schmidt, 1923, Bull. Amer. Mus. nat. Hist. 

49: 81, fig. 4. Type locality: Niapu, Belgian Congo [= D.R.C.]; 

Witte, 1941:202. 
Rhamnophis aethiopissa elgonensis Loveridge, 1929, Bull. U. S. 

natn,. Mus. 151: 24. Type locality: Yala (= Lukosa) River at the 

foot of Mount Elgon, Kenya; 1944: 129. 
Rhamnophis aethiopissa aethiopissa Loveridge, 1944: 126; Perret, 

1961: 136; Villiers. 1966: 1739; Stucki-Stirn, 1979: 335, figs. 
Rhamnophis aethiopissa ituriensis Loveridge, 1944: 128; Laurent. 

1956: 189, 355; 1960: 47 & 1964: 108; Bourgeois, 1968: 109, fig. 

43^6; Broadley, 1991:532. 
Thrasops aethiopissa elgonensis Loveridge, 1957: 264; Pitman, 

1974: 101, PI. T, fig. 3; Spawls, 1978: 5. 
Thrasops aethiopissa aethiopissa Hughes & Barry, 1969: 1018; 

Trape & Roux-Esteve, 1995: 40. 
Thrasops (Rhamnophis) aethiopissa Hinkel, 1992: 144, PI. 130. 

Diagnosis. Dorsal scales in 15-17 (rarely 19) rows at midbody, 
the vertebral row enlarged; ventrals 154-179; cloacal divided; 
subcaudals 1 17-159; two or three labials in contact with the lowest 
postocular: two large occipitals. 

Description. Supralabials 6-9, the 3 rd & 4 ,h . 4 ,h & 5 ,h or 5 th & 6 th 
entering orbit; infralabials 7-11, the first 3-6 in contact with the 
anterior sublinguals; preocular 1 (very rarely 2); postoculars 2-3 
(very rarely 4); a single temporal; two large occipitals (one longitu- 
dinally divided and the other semidivided in NMZB-UM 2548). 
Dorsals smooth, or vertebral and paravertebral rows keeled (Perret, 
1961) in 15-17 (very rarely 13 or 19) rows at midbody (usually 17 
rows in West Africa, Cameroon, Gabon and Central African Repub- 
lic, 15 rows elsewhere); ventrals 154-179; cloacal divided; 
subcaudals 1 1 7- 1 59 pairs, the lowest counts in Uganda and western 
Kenya. 



REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 
THRASOPS 



65 




J 00 400 tOO 100 '000 I TOO 1*00 \*M HOP WOO KJLOMITRfS 

I i 1 i 5 5 5 \ =^=\ 



RHAMNOPHIS & XYELEDONTOPHIS 




RHAMNOPHIS 
■ AETHIOPISSA 

• BATESII 
A AETHIOPISSA & BATESII 

♦ XYELEDONTOPHIS ULUGURUENSIS 

\ I 



100 O JOO *00 tOO iOO IO00 HOP 1*00 HOP HOP JOOO KH.OMCTMS 

Jiiim i r v -\ \ \ \ \ \ \ 5 v 

Fig. 5 Distributions of the genera Thrasops (upper), Rhamnophis and Xyelodontophis (lower). 



Coloration in life. Above, head olive-brown, uniform or poste- 
rior shields black-edged; body green with scales tipped or bordered 
with black; tail black with a green stripe on each scale row. Chin and 
throat yellow-green, rest of venter pale green, a dark median line on 
the subcaudals. 



SIZE. Largest 8 (MRAC 1 2257 - Kiroziret Forest, Kivu, D.R.C.) 
948 + 509 = 1457 mm (Laurent, 1956); largest 9 (NHRM 1979 - 
Bibindi, Cameroon, syntype of T. splendens) 950 + 520 = 1470 mm 
(Andersson, 1901). 



66 



D.G. BROADLEY AND V. WALLACH 



Habitat. Rain forest and gallery forest from sea level up to 2000 
metres. 

Distribution. Guinea east to the Democratic Republic of the 
Congo, Rwanda, Uganda and western Kenya, south to northern 
Angola and northwestern Zambia (Broadley, 1991) (Fig. 5).. 

REMARKS. Roux-Esteve (1965) placed R. a. ituriensis in the syn- 
onymy of the typical form and R. a. elgonensis hardly warrants 
subspecific recognition. Both subspecies were based on variable 
characters: the number of midbody scale rows, subcaudals and 
supralabials, and there are no major breaks in the distribution of the 
species. 

Rhamnophis batesii (Boulenger) 

Spotted Dagger-tooth Tree Snake 

Thrasops batesii Boulenger, 1908, Ann. Mag. nat. Hist. (8) 2: 93. 
Type localities: Akok and Efulen, Cameroon; Trape & Roux- 
Esteve, 1995: 40; Chippaux, 1999: 99. 

Rhamnophis batesii Schmidt, 1923: 83, fig. 5; Loveridge, 1944: 
125; Laurent, 1956: 355, PI. xx, fig. l;Perret, 1961: 1 36; Villiers, 
1966: 1739; Stucki-Stirn, 1979: 339. 

Diagnosis. Dorsal scales in 13 rows at midbody, vertebral row 
enlarged; ventrals 163-179; cloacal entire; subcaudals 92-1 19; two 
labials in contact with lowest postocular. 

Description. Supralabials 7 (rarely 6 or 8), the fourth & fifth 
(rarely third & fourth or fifth & sixth) entering orbit; infralabials 8 or 
9, the first 4-6 in contact with anterior sublinguals; preocular 1 
(rarely 2); postoculars 3 (rarely 2 or 4), 2 labials in contact with the 
lowest; a single temporal; 4 occipitals (3 in MRAC 19070 due to 
fusion of right hand pair; the median pair transversely divided in 
NMZB 13206). Dorsals smooth in 13-13-11 or 13-13-9 rows, verte- 
bral row enlarged; ventrals 163-179; cloacal entire; subcaudals 
92-123 pairs. 

COLORATION IN LIFE. Dorsum pale violet-brown, many scales 
black at the base and along lower edge, giving a plaited effect to the 
supracaudals. Chin and throat cream, rest of venter pale green, with 
black labial sutures and numerous black spots or blotches on the 
venter. 

SIZE. Largest 3 (MCZ 38393 - Batouri District. Cameroon) 827 + 
390 = 1217 mm; largest 9 (BMNH — ) 1450 + 350 = 1800 mm. 

Habitat. Rain forest between 400 and 1000 metres. 

DISTRIBUTION. Cameroon, Gabon and Congo-Brazzaville, east 
through the Congo basin to the Orientale and Kivu Provinces of the 
D.R.C. (Fig. 5). 

Xyelodontophis gen. nov. 

Diagnosis. A member of the tribe Dispholidini, differing from 
the other genera in the development of strongly curved rear maxil- 
lary teeth, which have sharp flanges anteriorly and posteriorly and 
narrow at the base, hence the name Xyelodontophis = Dagger-tooth 
Snake. Both species of Rhamnophis also have 'dagger-shaped' rear 
maxillary teeth, but they are less well developed and the teeth taper 
from base to tip, while Thelotornis and Dispholidus have large 
deeply grooved rear fangs. The new genus agrees with Thrasops and 
Thelotornis in having a shallowly forked ectopterygoid bone, whereas 
Rhamnophis and Dispholidus have a deeply forked ectopterygoid. 
In general form and scalation the new snake agrees with Thelotornis, 
but it lacks the distinctive horizontal pupil of that genus. 



Xyelodontophis uluguruensis sp. nov. 

Dagger-tooth Vine Snake 

HOLOTYPE. NMZB 7443 (Figs If, 2a, & 4a) an adult female from 
Lupanga Peak, Uluguru Mountains, Tanzania (06° 52' S: 37° 43' E), 
collected by Jon Lovett in November, 1983 (KMH 2636). Named 
for the Uluguru Mountains, to which it is probably endemic. 

PARATYPE. ZMB 48 1 53, an adult male from Bondwa Peak, Uluguru 
Mountains (06° 54 S: 37° 40' E) at 1 650 m, collected by D. Emmrich 
(DE413) in November, 1989. 

DIAGNOSIS. As for Xyelodontophis gen. nov. 

Description (paratype variations in parentheses). Rostral feebly 
recurved onto upper surface of snout; very large nostril in a single 
nasal; loreals 2; preocular 1; postoculars 3; temporals 1 +3; a pair of 
large occipitals behind the temporals, separated by elongate interpa- 
rietal; supralabials 8, the fourth and fifth entering the orbit; infralabials 
9, the first four or five in contact with the anterior sublinguals. 
Dorsals elongate, narrow, in 21-19-13 rows, moderately to feebly 
keeled, with single large apical pits; ventrals angular, but not keeled, 
168 (169); cloacal longitudinally divided; subcaudals 132+ (18+), 
tail truncated. The paratype male has an umbilical scar on ventrals 
147-149. 

Coloration in preservative. Top of head brown, shields nar- 
rowly margined with black, labials, chin and throat immaculate. 
Body grey-brown, bases of scales (and interstitial skin anteriorly) 
black; venter uniform pale grey apart from some irregular brown 
margins to the free edges of the ventrals. The paratype male has the 
head and nape bronze, supralabials immaculate yellow, rest of 
dorsum black speckled with green and brown in life (Emmrich, pers. 
comm.); chin and throat yellow, rest of venter rapidly darkening to 
black with a few light markings. 

Visceral anatomy (Tables 2^1). Umbilical scar-vent interval 
11.6% VS (10.7-12.5%); peritoneum black; hyoid posterior tip 
7.8% (7.6-8.1%); heart short 2.3% (2.0-2.6%), midpoint 25.2% 
(24.9-25.5%), junction of systemic arches ventrolateral and 0.71% 
(0.69-0.74%) heart length posterior to heart, right arch 0.33 diameter 
of left at junction; heart-liver gap 5.7% (5.5-5.9%), heart-liver 
interval 35.8% (34.3-37.2%); liver long 27.8% (26.4-29.2%) and 
narrow, midpoint 46.0% (45.7^16.2%), nearly contacting the gall 
bladder (liver-gall bladder gap 0.1% [0-0.2%]); liver-gall bladder 
interval 29.3% (28.3-30.4%), liver-kidney gap 28.0% (27.6-28.4%), 
liver-kidney interval 65.5%, liver length/right lung length ratio 
0.40; gall bladder 1.4% (1.2-1.6%), midpoint 60.7% (59.9-61.5%), 
located anterior to the subequal pancreas 1.7% (1.5-1.9%) with 
small spleen (0.7%) attached cranially; gall bladder-gonad gap 
10.7% (9.7-11.8%), gall bladder-gonad interval 23.0% (22.8- 
23.2%); gonads light yellow in color, right testis 4.8% (MP = 
74.4%), left testis 4.1% (MP = 79.9%), total testis midpoint 77.4%; 
right ovary 5.3% (MP = 74.9%) with 6 small ova and 7 follicles, left 
ovary 4.9% (MP = 81.6%) with 5 small ova and 8 follicles, total 
ovary midpoint 78.0%, total gonad midpoint 77.7%; gonad-kidney 
gap 4.8% (4.4-5.3%); adrenal glands orange, very narrow and 
elongate, adjacent to posterior end of gonads, right adrenal 2.3% 
(2.2-2.5%), midpoint 75.4% (74.8-76.1%), left adrenal 2.6% (2.5- 
2.7%), midpoint 81 .6% (8 1 .2-82. 1%), total adrenal midpoint 78.5% 
(78.4-78.6%); gall bladder-kidney gap 26.4% (26.3-26.5%), gall 
bladder-kidney interval 37.5% (36.3-38.8%); kidneys dark brown, 
segmented but compact with deep creases, right kidney 9.2% (8.0- 
10.4%) with 20 segments, midpoint 92.4%, right kidney length/liver 
length ratio 0.34; left kidney 7.0% (6.1-7.9%) with 21 segments, 



REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



67 



Table 2 Adult Dispholidini compared with Philothamnini: visceral characters as % snout-vent length 



GS 


n 


U-V 


Hy 


HLG 


HLI 


L 


LMP 


LGBG 


LGBI 


GBMP 


GBGG 


GBKG 


TGMP 


GKI 


TA 


TAMP 


TMP 


RL 


PT 


LL 


PA 


55 


8.6 


8.1 


7.3 


29.0 


18.8 


42.8 


3.5 


24.2 


56.6 


11.7 


23.9 


79.7 


23.7 






13.7 


76.9 


95.5 


1.4 


HL 


24 


7.9 


14.6 


9.0 


29.0 


17.3 


45.6 


10.3 


29.3 


65.3 


10.1 


20.2 


79.8 


22.1 






14.6 


76.1 


98.1 


1.4 


HS 


30 


6.8 


11.4 


8.8 


29.3 


17.7 


46.6 


5.6 


29.5 


66.2 


17.0 


22.1 


83.2 


20.0 


1.2 


84.7 


15.1 


78.4 


97.9 


0.8 


TF 


5 


8.2 


7.7 


8.8 


25.2 


14.2 


45.6 


6.2 


22.6 


60.0 


11.3 


23.3 


77.4 


24.4 


3.2 


77.9 


15.5 


68.4 


97.5 


2.6 


TJ 


15 


8.9 


7.2 


8.7 


26.6 


15.5 


44.5 


7.7 


20.0 


52.6 


9.8 


23.4 


76.1 


26.3 


5.3 


74.0 


14.6 


69.5 


96.4 


2.9 


TO 


2 


7.8 


6.7 


9.9 


26.0 


14.0 


43.9 


7.5 


23.0 


59.2 


8.8 


25.1 


76.1 


28.2 


2.6 


75.8 


14.1 


71.9 


97.6 


5.3 


TS 


2 


9.2 


10.7 


9.3 


26.9 


14.8 


44.8 


9.0 


25.0 


61.7 


7.3 


20.8 


76.0 


27.2 


4.6 


77.0 


14.8 


68.6 


96.4 


3.9 


RA 


12 


7.4 


10.2 


7.4 


27.1 


17.4 


44.0 


5.3 


24.4 


58.8 


12.5 


27.3 


79.0 


25.3 


3.9 


79.6 


14.5 


69.3 


96.3 


1.2 


RB 


2 


9.0 


9.9 


7.6 


30.0 


20.1 


45.1 


7.1 


28.4 


62.8 


13.4 


23.6 


79.6 


20.4 


2.9 


80.4 


14.3 


71.4 


97.7 


1.4 


DT 


12 


10.1 


9.0 


7.3 


29.0 


19.0 


44.1 


7.2 


27.5 


61.7 


7.7 


18.8 


77.0 


35.1 


2.9 


69.7 


14.2 


70.5 


96.3 


2.3 


DP 


1 


9.6 


6.4 


6.0 


30.2 


21.7 


42.9 


5.5 


28.5 


59.9 


12.1 


27.5 


79.5 


27.6 


4.9 


77.5 


13.6 


71.8 


96.6 


1.3 


TC 


4 


11.4 


8.2 


7.1 


31.9 


22.3 


44.2 


7.8 


31.8 


64.0 


8.2 


18.6 


78.6 


23.4 


3.0 


79.9 


13.6 


71.0 


96.1 


1.3 


TK 


13 


9.9 


7.1 


7.2 


32.2 


22.8 


47.8 


6.4 


31.0 


66.5 


10.5 


21.6 


82.7 


20.2 


2.2 


83.8 


15.2 


67.2 


95.7 


1.1 


TM 


5 


9.4 


7.2 


6.9 


36.3 


26.9 


45.7 


4.5 


33.3 


64.6 


9.3 


19.3 


80.5 


21.9 


2.5 


81.6 


13.4 


72.5 


96.9 


1.5 


TU 


3 


12.5 


7.2 


6.5 


34.0 


25.5 


44.9 


6.0 


33.2 


64.5 


10.5 


20.8 


80.5 


20.8 


1.8 


81.8 


13.4 


71.1 


96.0 


1.0 


XU 


2 


11.6 


7.8 


5.7 


35.8 


27.8 


46.0 


0.1 


29.3 


60.7 


10.7 


26.4 


77.7 


25.4 


4.9 


78.5 


13.8 


68.6 


93.4 


1.4 



(GS = genus/species: PA = Philothamnus angolensis, HL = Hapsidophrys lineatus, HS = Hapsidophrys smaragdinus, TF = Thrasops flavigularis, TJ = Thrasops jacksonii, TO 
= Thrasops occidentalis, TS = Thrasops schmidti, RA = Rhamnophis aethiopissa, RB = Rhamnophis batesii; DT = Dispholidus typus, DP = Dispholidus 'pemba', TC = 
Thelotornis capensis, TK = Thelotornis kirtlandii, TM = Thelotomis mossambicanus, TU = Thelotornis usambaricus, XU = Xyelodontophis uluguruensis; n = sample size. UV 
= umbilical scar to vent as c k total ventrals, Hy = hyoid posterior tip. HLG = heart-liver gap. HLI = heart-liver interval. L = liver length. LMP = liver midpoint, LGBG = liver- 
gall bladder gap. LGBI = liver-gall bladder interval, GBMP = gall bladder midpoint. GBGG = gall bladder-gonad gap. GBKG = gall bladder-kidney gap. TGMP = total gonad 
midpint. GKI = gonad-kidney interval, TA = total adrenal length, TAMP = total adrenal midpoint, TMP = trachea midpoint, RL = right lung length, PT = right lung posterior 
tip, LL = left lung length). 



Table 3 Adult Dispholidini compared with Philothamnini: visceral characters as ratios 



GS 


R/LSA 


K/L 


RK/L 


LK/RK 


KOL 


NTR 


AL/LL 


RB/LL 


LB 


SS/DV 


V/S 


LL/RL 


LW/LL 


PA 


0.39 


0.39 


0.37 


0.83 


0.76 


72 


0.54 


1 .00 


2.83 


0.19 


0.18 


0.02 


0.42 


HL 


0.36 


0.68 


0.67 


0.83 


0.80 


75 


- 


0.41 


0.53 


0.14 


0.17 


0.02 


0.30 


HS 


0.39 


0.49 


0.47 


0.85 


0.77 


101 


- 


0.23 


0.56 


0.13 


0.16 


0.01 


0.42 


TF 


0.33 


0.88 


0.66 


0.83 


0.38 


90 


0.31 


0.13 


0.60 


0.09 


0.25 


0.04 


0.21 


TJ 


0.41 


0.70 


0.57 


0.75 


0.40 


92 


0.36 


o.:i 


0.67 


0.10 


0.23 


0.04 


0.22 


TO 


0.37 


0.86 


0.71 


0.75 


0.45 


120 


0.24 


0.10 





0. 1 1 


0.24 


0.08 


0.13 


TS 


0.20 


0.93 


0.82 


0.78 


0.50 


89 


0. 1 3 


0.11 


0.50 


0.10 


0.26 


0.06 


0.29 


RA 


0.29 


0.48 


0.50 


0.69 


0.48 


76 


0.87 


0.24 


1.45 


0.10 


0.16 


0.02 


0.26 


RB 


0.22 


0.48 


0.46 


0.81 


0.76 


79 


1.00 


0.19 


1.67 


0.13 


0.15 


0.02 


0.35 


DT 


0.24 


0.65 


0.54 


0.78 


0.36 


79 


0.76 


0.80 


1.29 


0.17 


0.20 


0.03 


0.29 


DP 


0.25 


0.41 


0.33 


0.74 


0.40 


99 


0.90 


0.21 


4.00 


0.21 


0.15 


0.02 


0.30 


TC 


0.32 


0.59 


0.50 


0.68 


0.52 


71 


0.57 


0.21 


1.00 


0.13 


0.15 


0.02 


0.28 


TK 


0.26 


0.37 


0.37 


0.74 


0.55 


75 


0.54 


0.19 


2.7 


0.14 


0.15 


0.01 


0.25 


TM 


0.31 


0.45 


0.41 


0.72 


0.57 


76 


0.68 


0.20 


2.75 


0.16 


0.14 


0.02 


0.24 


TU 


0.38 


0.43 


0.39 


0.79 


0.63 


71 


0.79 


0.17 


3.00 


0.14 


0.12 


0.01 


0.39 


XU 


0.33 


0.35 


0.34 


0.76 


0.67 


59 


0.83 


0.42 


1 .00 


0.39 


0.13 


0.02 


0.20 



(genus/species acronyms as for Table 2, R/LSA = right systemic arch diameter/left systemic arch diameter. K/L = total kidney length/liver length. RK/L = right kidney length/ 
liver length, RA/RK = right adrenal length/right kidney length. LK/RK = left kidney length/right kidney length, KOL = kidney overlap/total kidney length, NTR = estimated 
number of tracheal rings/10% SVL, AL/LL = anterior lobe length/left lung length, RB/LL = right bronchus length/left lung length. LB = mean number of cartilages in left 
bronchus (including bronchial ring). SS/DV = semisaccular lung/dense vascular lung. V/S = vascular lung/saccular lung. LL/RL = left lung length/right lung length. LW/LL = 
left lung width/left lung lensUh). 



midpoint 94.0%, left kidney length/liver length ratio 0.26; left 
kidney/right kidney 0.76, kidney overlap 0.67; kidney-vent interval 
12.2% (1 1.6-12.8%), kidney-vent gap 2.5% (2.1-2.9%). 

Trachea 25.1% (25.0-25.2%) with an estimated 149 rings (144- 
154) or 58.9 (58.6-59.2) per 10% SVL, trachea midpoint 13.8% 
(13.7-13.9%), tracheal rings narrow and well-separated from their 
neighbours, lacking free tips, tracheal membrane expanded to 3.5 
times the tracheal ring circumference; tracheal lung lacking, only 
slight development of a cardiac lung (0.7%) anterior to the right 
lung; tracheal entry into right lung subterminal; right bronchus 0.3% 
with 3 cartilages; anterior lobe of right lung very short 1.0% (0.9- 
1 .2%), its connecting orifice of moderate diameter; right lung 68.6% 
(66.0-71.2%), right lung midpoint 60.1% (59.1-61.2%), vascular 
portion of right lung 8.0% (7.2-8.7%) with three tiers of faveoli 
distributed dorsoventrally around inner lung circumference; vascular 



lung lacking midventral avascular strip; faveolar pattern consists of 
transverse ribs enclosing transverse rows of paired diamond-shaped 
faveoli; vascular lung with semisaccular or sparse/dense vascular 
portion ratio 0.39 (0.35-0.42); saccular (avascular) lung 59.6% 
(58.0-61.3%), vascular lung/saccular lung ratio 0.13 (0.12-0.14), 
posterior tip of lung 93.9% (91.7-96.2%). 

Left lung complex consists of an orifice at 26.3% (26.2-26.5%), 
a bronchus 0. 1 % with two rings in female (bronchus absent in male), 
and a vestigial lung 1 .4% ( 1 . 1-1.6%). The left lung, with a left lung/ 
right lung ratio of 0.02, supports a reticulated network of trabeculae 
and has a width/length ratio of 0.20 (0.15-0.25). 

Xyelodontophis, while resembling Thelotornis in external mor- 
phology, is distinct from the latter genus in a number of internal 
characters: heart-liver gap, heart-liver interval, liver length, gall 
bladder midpoint, gall bladder-kidney gap, total gonad midpoint, 



68 



D.G. BROADLEY AND V. WALLACH 





> 




D 




> 




Q 




m 




CO 




05 




H 




H 




oi 




H 




Z 




_J 




§ 




J 




W 




< 




J 




§ 




f- 




-J 




s 




2 




j 




j 




fc 




-j 




2 




CU 




§ 




H 




< 




H 




o 




\A 




O 




Q. 




2 




o 




H 




a 




H 








> 




x 




a 




w 


C/l 


m 


Q 


O 






U 




n 




H 




3 


CQ 


O 


O 




J 


Ti 




■_ 




<D 






0- 


'> 


§ 


t_ 


J 







~- 




e 

CO 


J 


CD 




E 




o 








OJ 


_1 


c 


I 


CD 




W) 




--— 




O 


a 


o 


-j 


Cfl 


X 


'S 




D. 




E 

o. 




U 




-r 


d 
e 


CU 








-O 




SI 




F 


a 



1-; Oj oo o 

r~ t> t— ' a\ o o <n 



o o o o o o o 



O O n (N ^ - a 
00 — lO — O On -3" 
-M NtN N^ -h 



00 f^l tN \0 -" C\ ^ 
(N {N OJ n (N fO X 
©'©'©'© O © O 



O O O O O © 



© O © © O O © 



cn r- oo m cn — -^t 

in l^ \o in m ^ n 
o ©'©'©' © o © 



© — < © (N <N (N ■* 

^ — rn -- <N — < — ' 



ooo»im»iOt> 



in in ^ lo in ^t ■* 



tNMoor^ino; 
— < en en ^t- <n "*' 



»n oo oo m m no oo 

— — <N ■>** O -5t ■* 



r-; O) r~; C~> <N © t-; 

on © no r^ r~ <-« l-~ 
r- oo r- r- r- oo r- 



in cn r- r~ no © no 

oo \d rn rn mi \d c> 



I— ^ on m O m no CN 



oo cm — oi no m ■* 



(N -3" •<* On no O CO 



no — ro tj- r- -3- o 



in on r-; in q on r 
K oo oo t^ (^ \b in 



m co oo — on oo oo 



(N N "t (M (N * — 



.C >? i- J= •- J3 >> 



o $2 " 

d "3 H > 

x ° x- e 

.9- ii s$ 

S — ~ "« 

q. ~ x <l> 

MM s 

f ^ >,•* 

II J= <u o 



>, oo 



-o <U 



**' C -r- — 
W O U t3 

#.?" § 

«2^TI 
B O^ O 

i n*l 

cS £ a OS 

5 S c H 

»= •" • - -J 

■- u — ■-* 

g > aot/3 

to-g «-° 

fe a 3 a 

■S " <s .s 
= ^ ii s 

II . J -C 
O rt "^ 2 

e »d," 
S v'z ° 
S-S-g I 
ii A to a 

,_ (U o e 

O -O Q. C 

• • ^ an ^ 

^ — C u 






at 



i "O 



§•0 

o -J 



^1 

■ —' J£ 
'. = u 

: 'o « «■ 

■ •£ ii £ 

« -> > 

u Pi _ 



H 15 — _ 
;£ xi -a u 

M «2c 



S- c — — 



to 
a 


u 


c 


g 


_c 


II 

H 


> 


II 

< 


-5 


3 


II 


H 

- 

00 


z 


II 
> 


t^ 


~ 


II 


D 


g, 


c 


— 


« 
5 


_3 


:§ 


<u 


-2 


X 


_^ 


S. 


^ 


•CI 

eg 


c 


3 


% 


g 





<u 


S 


ii 


CQ 

cj 


SI) 


o 


> 




X 


II 


> 






II 


G 


1 


u 

•B 


3 

5 


— 

s 

60 




s 

o 


on 

c 

CO 


"5b 

c 

J3 




<U 


^0 


-3 


3 



S-iE2 



"S "° 



2 e 






REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



69 



gonad-kidney interval, total adrenal length and midpoint, adrenal/ 
kidney length ratio, total kidney/liver length ratio, number of tra- 
cheal rings, right bronchus/left lung ratio, semisaccular/dense 
vascular lung ratio, right lung posterior tip, left bronchus cartilages, 
and left lung width/length ratio. Xyelodontophis is also unique in 
differing from the rest of the Dispholidini in the means of the 
following characters: heart-liver interval, liver length, liver mid- 
point, liver-gall bladder gap, total gonad length, liver-kidney interval, 
gall bladder-kidney gap, total gonad length, right kidney and total 
kidney/liver length ratios, number of tracheal rings, semisaccular/ 
dense vascular lung ratio, lack of ventral avascular strip in vascular 
lung, liver/right lung length ratio, and posterior tip of right lung. 

In contrast to other Dispholidini, Xyelodontophis is most similar 
to Thelotomis in hyoid length, liver-gall bladder interval, and 
trachea midpoint; it is most similar to Rhamnophis and Thelotomis 
in total kidney midpoint, kidney-vent interval, and left lung length; 
it is most similar to Dispholidus, Rhamnophis and Thelotomis in 
total gonad midpoint; and it is most similar to Dispholidus in total 
adrenal length and adrenal/kidney length ratios. 

HEMIPENIS. In the paratype male the single organs extend to the 
seventh subcaudal, the sulcus spermaticus is simple (on the left 
organ the sulcus lies on the medial side, on the right organ it lies on 
the lateral surface). In situ, the basal portion (2 subcaudals) is nude 
with four large hooks, the medial portion (2 sc) is spinose and the 
apical portion (2 sc) is calyculate. The sulcus is bordered by two 
basal hooks 5 mm long and the two largest hooks (7 mm) are on the 
asulcate side. The calyces on the apex are smooth and form a 
network (with 1-2 mm cavities) very similar to the faveoli of the 
snake lung. The proximal 2-4 calyces are spinose with several tiny 
spinules on each calyx. The spines completely surround the organ 
and are arranged in 7-8 rows, increasing in size from distal (4 rows, 
1 mm long) through medial (3 rows, 2 mm) to proximal (1 row, 3 
mm); there are 7 large spines on the right organ and 8 on the left. The 
everted organ would probably show some resemblance to that of 
Thelotomis kirtlandii (Doucet, 1963: Fig. 40). 

Size. Length 740 + 407+ mm (snout-vent 830 mm, tail truncated 
near base). 

DIET. The holotype contained a recently swallowed leaf chame- 
leon, Rhampholeon uluguruensis, an endemic species recently 
described from Bondwa Peak (Tilbury & Emmrich, 1996). 

Habitat. Montane evergreen forest. The habitat is described by 
Tilbury & Emmrich (1996). 

Distribution. Probably endemic to the Uluguru Mountains (Fig. 
5). 

Thelotomis kirtlandii (Hallowell) Forest Vine Snake 

Leptophis Kirtlandii Hallowell, 1844, Proc. Acad. not. Sci. Phila- 
delphia: 62. Type locality: Liberia, type ANSP 5271. 

Oxybelis Lecomtei Dumeril & Bibron, 1854, Erpet. Gen., 7: 821. 
Type locality: Gabon. 

Tragophis rufulus Dumeril & Bibron, 1854, Erpet. Gen., 7: 827. 
Type locality: Senegal. 
\ Oxybelis violacea Fischer, 1856, Abhand. Nat. Ver. Hamburg, 3: 91, 
PI. ii, fig. 7. Type locality: Edina, Grand Bassa County, Liberia. 

Dryiophis Kirtlandii Bocage, 1895: 119 (part). 
I Thelotomis kirtlandii Schmidt, 1923: 11 2, PI. xiv; Bogert, 1940:69; 
Witte, 1953: 247, fig. 82; Laurent, 1964: 116. 

Thelotomis kirtlandii kirtlandii Loveridge, 1944: 149 (part). 



neck with black crossbands; supralabials immaculate or with fine 
green or grey stipple; rostral and nasals strongly recurved onto top of 
snout; infralabials 7-11 (mode 9); ventrals 162-189; subcaudals 

132-172. 

Description. Rostral and anterior nasals recurved onto top of 
snout; a single loreal (in eastern populations); preocular 1 ; postoculars 
3 (2 in two specimens from Digba through fusions with supraocular 
or fifth labial) ;temporals 1+2 (very rarely 1 + 1 or 2+2); supralabials 
8 (rarely 9 or 10), the fourth and fifth (rarely fifth and sixth) entering 
the orbit: infralabials 7 to 1 1 , the first 4 or 5 (very rarely 3) in contact 
with the anterior sublinguals. Dorsal scales feebly keeled in 19-19- 
13 rows (17 rows at midbody in four specimens from Kivu: Laurent, 
1956, 1960); ventrals 164-179in 66, 164-189 in99;cloacal divided; 
subcaudals 135-157 in 66, 138-165 in99. 

COLORATION. Top of head uniform green, supralabials white, 
often with fine green or grey stipple; body mottled grey, green and 
brown, with black crossbars anteriorly (ZMUC R631282 lacks 
black markings on the neck), lighter below. The specimen illustrated 
by Hinkel ( 1992: fig. 129) appears to be uniform dark brown on top 
of the head, with heavy brown infuscation on the labials. This could 
be a captive specimen that has been exposed to strong sunlight, such 
a change has been observed in a captive Thelotomis at Watamu on 
the Kenya coast (S. Spawls, pers comm.). 

SIZE. Largest 6 (AMNH 12279 - Niangara, D.R.C.) 850 + 480 = 
1 330 mm, largest 9(ZMUC R63 1 282 - Massisiswi, Udzungwa Mts, 
Tanzania) 1050 + 660 = 1710 mm. 

Habitat. Lowland forest in west and central Africa, relict popul- 
ations in montane forests in Tanzania. 

DISTRIBUTION. Islands of the Bijagos Archipelago, Guinea Bissau, 
east through forested areas of west Africa and the Congo basin to 
Uganda and southern Sudan, south to northern Angola, northwest- 
ern Zambia (Broadley, 1991 ) and south-central Tanzania (Rasmussen, 
1997) (Fig. 6). 

Thelotomis usambaricus Broadley Usambara Vine Snake 

Thelotomis kirtlandii (not Hallowell) Stejneger, 1893: 733. 

Thelotomis kirtlandii kirtlandii (not Hallowell) Loveridge, 1944: 
149 (part). 

Thelotomis capensis mossambicanus (not Bocage) Broadley, 1979: 
126 (part); Rasmussen, 1997: 138 (part). 

Thelotomis usambaricus Broadley, 2001, Afr. J. Herpetol. 50 (2): 
58. Type locality: Amani Nature Reserve, (Kwamkoro/ 
Kwemsambia Forest Reserve), East Usambara Mountains, Tan- 
zania. Holotype: NMZB 16182 

DIAGNOSIS. Top of head, including temporal region, uniform green; 
neck with black chevrons; supralabials with scattered black spots, 
usually including a triangle on the sixth labial; rostral and nasals not, 
or only feebly, recurved onto top of snout; infralabials 9-13 (mode 
11); ventrals 145-169; subcaudals 143-175. 

Description. Rostral just visible from above; nasal entire; loreals 

I or 2; preocular 1 ; postoculars 3; temporals 1 + 2 (very rarely 1 + 3); 
occipitals 2, separated by a small interoccipital; supralabials 8 (very 
rarely 9), the fourth and fifth or third, fourth and fifth entering the 
orbit; infralabials 9 to 13, the first 4 or 5 in contact with the anterior 
sublinguals. Dorsal scales very feebly keeled, in 19-19-13 or 19-19- 

II rows; ventrals 156-166 in 66, 145-169 in 99; cloacal divided; 
paired subcaudals 146-175 in 66, 143-169 in 99. 



Diagnosis. Top of head, including temporal region, uniform green; Coloration. Top of head, including temporals, uniform green in 



70 



D.G. BROADLEY AND V. WALLACH 



life, supralabials, chin and throat white or pale orange, with a few 
black spots and usually a speckled black triangle extending back 
from the eye through the lower postocular and sixth labial to the lip, 
a few black spots on posterior sublinguals and gulars; dorsum 
mottled brown, green and pale grey, three or four vague black 
chevrons on neck (more distinct in subadults); venter mottled pale 
brown and green. 

Size. Largest 6 (BMNH 1974.547) 640 + 454 = 1094 mm; 
largest 9(ZMUC R63 13 10) 790 + 490 = 1280 mm, both from Amani. 

Habitat. Coastal forest. 

DISTRIBUTION. The Usambara Mountains, with apparently relict 
populations on the lower slopes of other isolated mountains in the 
Eastern Arc chain and on the Kenya coast (Fig. 6). 

Thelotornis mossambicanus (Bocage) Eastern Vine Snake 

Oxybelis Lecomtei (not Dumeril & Bibron) Peters, 1 854: 623 (part). 

Thelotornis Kirtlandii (not Hallowell) Peters, 1882: 131 (part), PI. 
xix, fig. 2 

Dryiophis Kirtlandii var. mossambicana Bocage, 1895, Herp. An- 
gola & Congo: 119. Type locality: Manica, Mozambique. 
Lectotype MBL 1843 (destroyed). 

Thelotornis kirtlandii capensis (not A. Smith) Mertens, 1937: 14. 

Thelotornis capensis (not A. Smith) Bogert, 1940: 70 (part), fig. 1 1 . 

Thelotornis capensis capensis (not A. Smith) Laurent, 1956: 230 & 
378. 

Thelotornis capensis mossambicanus Broadley, 1979: 129. 

Thelotornis mossambicanus Broadley, 2001: 60. 

Diagnosis. Top of head green to pale brown, uniform or speckled 
with black; temporals brown speckled with black; neck with black 
lateral blotch; supralabials with scattered black spots, including a 
triangle on the sixth labial; rostral and nasals not, or only feebly, 
recurved onto top of snout; infralabials 9-13 (mode 11); ventrals 
144-172; subcaudals 123-167. 

Description. Rostral and nasals barely visible from above; loreals 
usually 2 (rarely 1, very rarely or 3); preocular 1; postoculars 3 
(rarely 2 or 4); temporals 1 + 2 (very rarely 1 + 1, 1 + 3 or 2 + 2); 
supralabials 8 (rarely 9, very rarely 6 or 7), the fourth and fifth 
(rarely fifth and sixth, very rarely third and fourth, or third, or fifth 
only) entering orbit; infralabials 9-13, mode 11, the first 4 or 5 in 
contact with the anterior sublinguals; dorsal scales usually in 19-19- 
11 or 19-19-13 rows, very rarely 17,21 or 23 rows at midbody (23 
recorded by Rasmussen, 1997); ventrals 144-168 in 66, 145-172 
in??; cloacal divided; subcaudals 131-167 in 66, 123-153 in??. 

Coloration. Crown of head uniform green or with a black 
speckled Y-shaped marking, or brownish, entirely speckled with 
black (the two extremes may occur within a population, as on Mafia 
Island); temporal region always brown, speckled with black; 
supralabials white spotted with black, including a triangle on sixth 
labial, chin and throat speckled with black; dorsum ash grey with 
diagonal rows of whitish blotches and flecks of brown and pink or 
orange, neck with one or two elongate black blotches; venter 
greyish, streaked with brown. 

Size. Largest 6 (MHNG 1 376.34 -Newala, Tanzania) 9 10 + 525+ 
(tail truncated); largest? (NMZB-UM 4157 - Mutare, Zimbabwe) 
895 + 510= 1405 mm, but MCZ 18476 from Zengeragusu, Tanza- 
nia, has a snout-vent length of 920 mm (tail truncated). 

Habitat. Savanna and coastal forest. 

Distribution. Southern Somalia south to central Mozambique at 



about 22°30' S, west to the shores of Lake Tanganyika, Malawi and 
eastern Zimbabwe (Fig. 6). 

Thelotornis capensis capensis A. Smith 

Southeastern Savanna Vine Snake 

Thelotornis capensis A. Smith, 1 849, ///. Zool. S. Africa, Rept. App.: 
19. Type locality: 'Kaffirland and the country towards Port 
Natal', i.e. Durban (type lost). 

Thelotornis kirtlandii capensis Loveridge, 1944: 154 (part). 

Thelotornis capensis capensis Broadley, 1979: 126. 

Diagnosis. Top of head blue-green with pink and black speckling 
forming a 'Y' or 'T' marking, or speckling covering entire top of 
head; temporals pink margined with black; neck with black lateral 
blotches; supralabials with scattered black spots, including a trian- 
gle on the sixth labial; rostral and nasals not, or only feebly, recurved 
onto top of snout; infralabials 9-13 (mode 11); ventrals 144-164; 
subcaudals 127-155. 

Description. Rostral and nasals barely visible from above; loreals 
usually 2 (rarely 1, very rarely or 3); preocular 1; postoculars 3 
(rarely 2 or 4); temporals 1 + 2 (very rarely 1 + 1 or 1 + 3); 
supralabials 8 (very rarely 7 or 9), the fourth and fifth (very rarely 
third & fourth, fifth & sixth or third, fourth and fifth) entering orbit; 
infralabials 9-13, mode 11, the first 4 or 5 (very rarely 3 or 6) in 
contact with anterior sublinguals; dorsal scales usually in 19-19-13 
rows , rarely in 17 rows at midbody (15 rows only in TMP 45554); 
ventrals 144-160 in 66, 148-162 in??; cloacal divided; subcaudals 
133-155 in 66, 127-147 in??. 

Size. Largest 6 (NMZB 6389 - Gwanda, Zimbabwe) 830 + 506 = 
1336 mm; largest? (TMP 56 15 - Hectorspruit, Mpumalanga, South 
Africa) 91 1 + 455 = 1366 mm. 

Habitat. Savanna. 

Distribution. Southwestern Zimbabwe and southeastern Bot- 
swana, south through the northern provinces of South Africa and 
Swaziland to southern Mozambique and KwaZulu-Natal (Fig. 6). 

Thelotornis capensis oatesii (Giinther) 

(Dates' Savanna Vine Snake 

Oxybelis Lecomtei (not Dumeril & Bibron) Peters, 1854: 623 (part, 
Tete). 

Dryiophis oatesii Giinther, 1881, In Oates' Matabeleland and the 
Victoria Falls, App. : 330, Col. PI. D. Type locality: Matabeleland 
(= western Zimbabwe), type BMNH 1946.1.9.76. 

Thelotornis Kirtlandii (not Hallowell) Peters, 1882: 131 (part). 

Thelotornis kirtlandii capensis Loveridge, 1944: 154 (part). 

Thelotornis capensis (not A. Smith) Witte, 1953: 249, fig. 82. 

Thelotornis kirtlandii oatesii Loveridge, 1953: 277. 

Thelotornis capensis oatesii Laurent, 1956: 231, fig. 35 

Diagnosis. Top of head blue-green with pink and black speckling 
forming a 'Y' or 'T' marking; temporals pink margined with black; 
neck with black lateral blotches; supralabials with scattered black 
spots, including a triangle on the sixth labial; rostral and nasals not, 
or only feebly, recurved onto top of snout; infralabials 9-13 (mode 
11); ventrals 150-177; subcaudals 126-168. 

DESCRIPTION. Rostral and nasals barely visible from above; loreals 
usually 2 (rarely 1, very rarely 0); preocular 1; postoculars 3 (rarely 
2, very rarely 1 or 4); temporals 1 + 2 (very rarely 1 + 3 or 1 + 1); 
supralabials 8 (rarely 7, very rarely 9), the fourth and fifth (very 
rarely third & fourth, fifth & sixth, third, fourth & fifth, or third, or 






REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



71 




J o igojoojagigp kxp m» t«oo noo nop w oo ■jlowitris 
m i ni i i S i 5 5 5 5 v T 



Fig. 6 Distribution of the genus Thelotornis. 



72 



D.G. BROADLEY AND V. WALLACH 



fourth only) entering orbit; infralabials 9-13, mode 1 1 , the first 4 or 
5 (rarely 3) in contact with anterior sublinguals; dorsal scales 
usually in 19-19-11 or 19-19-13 rows, very rarely 17 rows at 
midbody; ventrals 150-177 in 33, 153-177 in 99; cloacal divided; 
subcaudals 132-173 in 33, 126-168 in 99 

SIZE. Largest 3 (NMZB 3 828 - Mtorashanga, Zimbabwe) 1 062 + 
620 = 1682 mm; largest 9 (NMZB 3600 - Lake Kariba, Zimbabwe) 
975 + 560 = 1535 mm, but NMZB-UM 1061 from Shurugwe, 
Zimbabwe, has a snout- vent length of 1050 mm (tail truncated). 

Habitat. Savanna. 

Distribution. Southern Angola and northern Namibia, west 
through northern Botswana, Zambia and southeast Katanga (D.R.C.) 
to Zimbabwe, western Mozambique and Malawi (Fig. 6). 



Dispholidus typus (A. Smith) 



Boomslang 



Bucephalus typus A. Smith, 1828, South African Commercial Ad- 
vertiser 3 ( 144): 2, col. 4. Type locality: Eastern districts of South 
Africa; 1829, Zool. Journ., 4: 441 (B. typicus). 

Bucephalus Jardineii A. Smith, 1828, South African Commercial 
Advertiser 3 (144): 2, col. 4. Type locality: Cape Town, South 
Africa; 1829, Zool. Journ., 4: 422. 

Bucephalus gutturalis A. Smith, 1828, South African Commercial 
Advertiser 2> (144): 2, col. 4. Type locality: Forests of the eastern 
districts of South Africa; 1829, Zool. Journ., 4: 442. 

Bucephalus Bellii A. Smith, 1828, South African Commercial Ad- 
vertiser '3 (144): 2, col. 4. Type locality: Eastern districts of South 
Africa; 1829, Zool. Journ., 4: 442. 

Dispholidus Lalandii Duvernoy, 1832, Ann. Sci. Nat. (Paris) 26: 
150. Type locality: Cape of Good Hope. 

Dendrophis colubrina Schlegel, 1837, Essai Phys. Serp. 2: 238, PI. 
ix, fig. 14-16. Type locality: Rondesbosch, [Western] Cape Prov- 
ince, South Africa. 

Bucephalus viridis A. Smith, 1838, lllus. Zool. S. Africa, Rept.: PI. 
iii. Type locality: Old Latakoo [Northern Cape Province], South 
Africa. 

Bucephalus capensis A. Smith, 1 84 1 , lllus. Zool. S. Africa, Rept. : PI. 
x-xiii. Type locality: Cape Province, South Africa; Bocage, 
1895: 121. 

Dendrophis pseudodipsas Bianconi, 1848, Nuovi Ann. Sci.Nat. (2) 
10: 108, PI. iv, fig. 2 & 1850. Spec. Zool. Mosamb. 40, PI. iv, fig. 
2. Type locality: [Inhambane] Mozambique. Holotype: Bologna 
100296. 

Thrasops jacksonii mossambicus Mertens, 1937, Abhand. 
senckenberg. naturf. Ges., No. 435: 13. Type locality: Cheringoma 
Farm, Inhaminga, Mozambique. Holotype SMF 22246. 

Dispholidus typus kivuensis Laurent, 1955, Revue Zool. Bot. Afr. 51: 
127. Type locality: Uvira, Kivu. Congo Beige [= D.R.C.]. Holotype 
MRAC 17505. 

Dispholidus typus punctatus Laurent, 1955, Revue Zool. Bot. Afr. 
51: 129. Type locality: Dundo, Angola. Holotype MRAC 17395. 

Dispholidus typus occidentalis Perret, 1961, Bull. Soc. neuchateloise 
Sci. nat. 84: 138. Type locality: Cameroon, no type designated. 

Diagnosis. Dorsal scales strongly keeled in 19 (rarely 17 or 21) 
rows at midbody; ventrals 164-201; anal divided; subcaudals 94- 
142. 

DESCRIPTION. Supralabials 7 (rarely 8 or 6), the third and fourth 
(rarely 5 th & 6 th ) entering orbit; lower labials 8-13, the first 3-6 in 
contact with anterior sublinguals; preocular 1 ; postoculars 3 (very 
rarely 2 or 4), the lower in contact with two labials; temporals 1 + 2 
(very rarely 1 +1, 1 +3, 2+ 1,2 + 2 or 2 +3); three enlarged 



occipitals, the middle one subtriangular. Dorsals strongly keeled in 
19 (rarely 17 or 21) rows; ventrals 164-201; cloacal divided; 
subcaudals 94-142 pairs. 

COLORATION IN LIFE. Juveniles are speckled dark grey-brown 
above, with paired blue spots on some adjacent scales that become 
visible when the skin is stretched, the lower scale rows are grey and 
the venter is white, heavily stippled with dark red-brown. The head 
is brown above, the labials and chin white, sometimes with some 
black spots, and the throat is bright yellow. The iris of the eye is 
bright green. The juvenile coloration is gradually lost as the snake 
approaches one metre in length and there is great variation in adult 
colour pattern. 

Males are usually green, with or without black-edged scales, 
females usually olive or brown above, paler below. This sexual 
dimorphism in colour pattern does not always apply, for example 
green females are not uncommon in Mozambique and KwaZulu- 
Natal, while in southwestern Zimbabwe some males are olive-brown 
above and duck-egg blue below. In East Africa a uniform black 
phase may occur in either sex. In the Eastern Cape Province (the 
'type locality') males are usually black above, each scale and head 
shield with a green or yellow spot, venter yellow-green, each ventral 
bordered with black, but in the southwestern Cape the dorsum is 
uniform black and the venter yellow. In the western form described 
as D. t. punctatus Laurent, the males are black above, each scale or 
head shield with an orange spot, ventrals violet edged with black. 
Females are usually red-brown above, paler below. 

Size. Largest 3 (NMZB 3947 - Mutoko, Zimbabwe) 1290 + 530 
= 1820 mm; largest 9 (NMZB 3820 - Makote, Newala, Tanzania) 
1447 + 475 = 1922 mm (tail tip truncated). C.J.P Ionides recorded a 
brown male from Tanzania that measured 2 1 34 mm (Pitman, 1 974). 

Habitat. Savanna. 

Distribution. Senegal east to the Horn of Africa, south to the 
southwestern Cape, excluding areas of rain forest, grassland and 
desert. 

Remarks. The data for the solitary specimen examined from 
Pemba Island, Tanzania (MCZ 45587), confirm the long held opin- 
ion of Barry Hughes that this population is taxonomically distinct: 
he will describe it when he has access to more material. The 
subspecies described by Laurent (1955) were based on male colora- 
tion and subcaudal counts, but there is clinal variation in both 
characters. The species needs to be reviewed, using material from 
throughout its extensive range. 

Key to the genera and species of Dispholidini 



la. 
lb. 

2a. 

2b. 
3a. 
3b. 
4a. 



Nasal divided; rear maxillary teeth not grooved 

Nasal entire; enlarged grooved fangs on posterior maxilla . 



Head elongate with two loreals in tandem; temporals 1 +2; maxillary 

teeth 17, the last three enlarged and dagger-shaped 

Xyelodontophis uluguruensis 

Head short with single loreal; temporals 1+1 or one only; maxillary 
teeth 20-38, the last three enlarged 3 

Vertebral scale row enlarged; a single temporal; 2 or 4 enlarged 
occipitals 4 

Vertebral scale row not enlarged; 1 + 1 temporals: no enlarged 
occipitals 5 

Cloacal shield entire; midbody scale rows 13; occipitals 4 

Rhamnophis batesii 






REVIEW OF DISPHOLIDINI WITH NEW TANZANIAN GENUS AND SPECIES 



73 



4b. Cloacal shield divided; midbody scale rows 15-19; occipitals 2 
Rhamnophis aethiopissa 

5a. Dorsal scales twice as long as lower row of laterals and ventrals; 
midbody scale rows 13 (rarely 15) Thrasops flavigularis 

5b. Dorsal scales subequal in length to the ventrals; midbody scale rows 
15-19 (rarely 21) 6 

6a. Three supralabials in contact with lower postocular 

Thrasops occidentalis 

6b. Two supralabials in contact with lower postocular 7 



7a. Midbody scale rows usually 19; ventrals 187-214 

Thrasops jacksonii 

7b. Midbody scale rows usually 17; ventrals 168-184 

Thrasops schmidti 

8a. Head elongate; pupil horizontal, keyhole shaped; eight supralabials, 
fourth and fifth entering orbit 9 

8b. Head short; pupil round or pear-shaped; seven supralabials, third 
and fourth entering orbit Dispholidus typus 

9a. Top of head, including temporal region, uniform green; black 
crossbands or chevrons on neck; habitat forest 10 

9b. Top of head uniform green or black speckled, temporal region 
always brown, speckled with black, or pink, shields margined with 
black; neck with black lateral blotches; habitat usually savanna 
1 1 

10a. Rostral and nasals strongly recurved onto top of snout; infralabials 
7-11, mode 9; supralabials immaculate or with fine green or grey 
stipple T kirtlandii 

10b. Rostral and nasals not, or only slightly, recurved onto top of snout; 
infralabials, 9-13, mode 1 1; supralabials with scattered black spots, 
usually including a triangle on the sixth labial T. usambaricus 

I la. Top of head bright green to pale brown, uniform or speckled 

with black; temporals brown speckled with black 

T mossambicanus 

I I b. Top of head blue-green with pink and black speckling forming a 'Y' 

or 'T' marking, or speckling covering entire top of head; temporals 
pink margined with black T capensis 



Acknowledgements. We are indebted to K.M. Howell (UDSM) for 
donating the snake that becomes the holotype of Xyelodontophis uluguruensis. 
VW wishes to particularly thank the following curators that permitted 
dissection of material in their care: K.-S. Chifundera (IRSL). G. Lenglet 
(IRSNB), J. Hanken and J.P. Rosado (MCZ) and R. Gunther (ZMB). DGB is 
grateful to D. Rotich (NMK) for the loan of Thrasops schmidti material and 
I. Ineich (MNHN), A. Resetar (FMNH), R. Gunther (ZMB) and G. Lenglet 
(IRSNB) for printouts of their Dispholidini holdings. 



REFERENCES 



Bocage, J. V. Barboza du 1895. Herpetologie cT Angola et du Congo, xx + 203 pp. 
Lisbonne: Imprimerie Nationale 

Bogert, C. M. 1940. Herpetological results of the Vernay Angola Ezpedilion. with 
notes on African reptiles in other collections. Part I. Snakes, including an arrange- 
ment of African Colubridae. Bulletin of the American Museum of Natural History 77: 
1-107. 

Boulenger, G. A. 1 896. Catalogue of the snakes in the British Museum (Natural 
History). 3: xiv + 727 pp. 

Bourgeois, M. 1968. Contribution a la morphologie comparee du crane des ophidiens 
de 1' Afrique Centrale. Publications de I'Universite Officielle du Congo a Lubumbashi 
18: 1-293. 

Broadley, D. G. 1966. A review of the Natal green snake. Philothanmus natalensis (A. 



Smith), with a description of a new subspecies. Annals of the Natal Museum 18 (2): 

417-423. 
1979. Problems presented by geographical variation in the African vine snakes of 

the genus Thelotornis. South African Journal of Zoology 14: 125-131. 
1983. Fit-Simons ' Snakes of southern Africa. 387 pp. Parklands: Jonathan Ball & 

Ad. Donker. 
1991. The herpetofauna of northern Mwinilunga District, northwestern Zambia. 

Arnoldia Zimbabwe 9 (37): 519-538. 
2001. A review of the genus Thelotornis A. Smith in eastern Africa, with the 

description of a new species from the Usambara Mountains (Serpentes: Colubridae: 

Dispholidini). African Journal of Herpetology 50 (2): 53-70. 
Cansdale, G. S. 1961. West African Snakes. London: Longmans. 74 pp. 
Chippaux, J. -P. 1999. Les serpents d 'Afrique occidentals et centrale. Paris: Editions 

de I'lRD (Collection Faune et Flore tropicales 35). 278 pp. 
Doucet, J. 1963. Les serpents de laRepubliquedeCoted'Ivoire./U/n Tropica 20: 201- 

340. 
Dowling, H. G. & Duellman, W. E. 1978. Systematic Herpetology: a synopsis of 

families and higher categories. 118 + vii pp. New York: HISS Publications. 
FitzSimons, V. F. M. 1 962. The snakes of southern Africa. 423 pp. Cape Town/ 

Johannesburg: Purnell. 
Hinkel, H. 1982. Herpetofauna. In: Fischer. E. & Hinkel. H. La Nature et I 'Environment 

du Rwanda. Mainz: Ministerium des Innem und fiir Sport. Rheinland-Pfalz. 452 pp. 
Hughes, B. 1983. African snake faunas. Bonner Zoologische Beitrage 34: 31 1-356. 
1985. Progress on a taxonomic revision of the African Green Tree Snakes 

(Philothanmus spp. ). Proceedings of the International Symposium on African 

Vertebates. Bonn. 19X5: 511- 530. 
Hughes, B. & Barry. D. H. 1969. The snakes of Ghana: a checklist and key. Bulletin 

de I' Inslitut Fondanieniul de V Afrique noire, ser. A, 31: 1004-1041. 
Jackson, K. & Fritts, T. H. 1995. Evidence from looth surface morphology for a 

posterior maxillary origin of the proleroglyph fang. Amphibia-Reptilia 16: 273-288. 
Kardong, K. V. 1979. 'Protovipers' and the evolution of snake fangs. Evolution ii: 

433-443. 
Kardong, K. V. & Lavin-Murcio, P. A. 1993. Venom delivery of snakes as high- 
pressure and low-pressure systems. Copeia 1993: 650-664. 
Kochva, E. 1978. Oral glands of the Reptilia. In Gans. C. & Cans. K.A. (eds) Biology 

of the Reptilia 8. New York: Academic Press. Pp. 43-94. 
Laurent, R. F. 1955. Diagnoses preliminaires de quelques Serpents venimeux. Revue 

Zoologique et Botanique Africaine 51: 127-139. 
1956. Contribution a I Herpetologie de la Region des Grands Lacs de l'Afrique 

Centrale. Annates du Musee Royal du Congo Beige. Ser. 8vo. 48: 1-390. 
1960. Notes complementaires sur les Cheloniens et les Ophidiens du Congo 

oriental. Annates du Musee Royal du Congo Beige. Ser. 8vo. 84: 1-86. 
Leviton, A. E„ Gibbs, R. H„ Jr., Heal, E. & Dawson, C. E. 1985. Standards in 

herpelology and ichthyology: Pari I. Standard symbolic codes for institutional 

resource collections in herpetology and ichthyology. Copeia 1985: 802-832. 
Loveridgt, A. 1923. Notes on East African snakes collected 1918-23. Proceedings of 

the Zoological Society of London: 871-897. 

1936. New tree snakes of the genera Thrasops and Dendraspis from Kenya 

Colony. Proceedings of the Biological Society of Washington 49: 63-66. 

1944. Further revisions of African snake genera. Bulletin of the Museum of 

Comparative Zoology 95: 121-247. 

1953. Zoological results of a fifth expedition to East Africa. Ill Reptiles from 

Nyasaland and Tete. Bulletin of the Museum of Comparative Zoology 110: 143-322. 

1957. Check list of the reptiles and amphibians of East Africa (Uganda; Kenya; 

Tanganyika; Zanzibar). Bulletin of the Museum of Comparative Zoology 117: 151- 
362, i-xxiv. 

Meier, J. 1981. The fangs of Dispholidus typus Smith and Thelotornis kirtlandii Smith 

(Serpentes: Colubridae). Revue Suisse Zoologique 88: 897-902. 
Mertens, R. 1937. Reptilien und Amphibien aus dem sudlichen Inner-Afrika. 

Abhandlungen der Senckengergischen Natuiforschenden Gesellschaft 435: 1—23. 
Peters, W. C. H. 1854. Diagnosen neuer Batrachier, welche zusammen mit der 

fruher (24. Juli und 17. August) gegebenen Ubersicht der Schlangen und Eide- 

schen mitgetheilt werden. Bericht iiber die zur Bekanntmachung geeigneten 

Verhandlungen der kdniglich-preussischen Akademie der Wissenschaften zu 

Berlin: 614-628. 
1882. Natunvissenschaftliche Reise nach Mossambique aufBefehl seiner Majesldt 

des Konigs Freidrich Wilhelm IV. In der Jahren 1842 bis 1848 ausgefuhrt von 

Wilhelm C. H. Peters. Zoologie III. Amphibien. 191 pp. Berlin: G. Reimer 
Perret, J.-L. 1961. Etudes herpetologiques Africaines III. Bulletin de la Societe 

neuchateloise des Sciences naturelles 84: 133-138. 
Pitman, C. R.. 1974. A guide to the snakes of Uganda. (Revised Edition) 290 pp. 

Codicote: Wheldon & Wesley. 
Rasmussen, J. B. 1997. Tanzanian records for vine snakes of the genus Thelotornis. 

with special reference to the Udzungwa Mountains. African Journal of Herpetology 

46: 137-142. 
Roux-Esteve, R. 1965. Les Serpents de la region de La Maboke-Boukoko. Cahiersde 

laMabokei: 51-92. 
Schmidt, K. P. 1923. Contributions to the herpetology of the Belgian Congo based on 



74 



D.G. BROADLEY AND V. WALLACH 



the collection of the American Museum Congo Expedition, 1909-1915. Bulletin of 

the American Museum of Natural History 49: 1-146. 
Spawls, S. 1 978. A checklist of the snakes of Kenya. Journal of the East Africa Natural 

History Society & National Museum 31 (167): 1-18. 
Stucki-Stirn, M. C. 1 979. Snake Report 721. A comparative study of the herpetological 

fauna of the former West Cameroon. 650 pp. Teuffenal: Herpeto-Verlag. 
Taub, A. M. 1967. Comparative histological studies on Duvernoy's gland of colubrid 

snakes. Bulletin of the American Museum of Natural History 138: 1-50. 
Tilbury, C. R. & Emmrich, D. 1996. A new dwarf forest chameleon (Squamata: 

Rhampholeon Giinther 1 874) from Tanzania. East Africa with notes on its infrageneric 

and zoogeographic relationships. Tropical Zoology 9: 61-71. 
Trape, J. F. & Roux-Esteve, R. 1995. Les serpents du Congo: liste commentee et cle 

de determination. Journal of African Zoology 109: 31-50. 
Underwood, G. 1967. A Contribution to the Classification of Snakes. British Museum 

(Natural History) Publication No. 653. Trustees of the British Museum (Natural 

History), London. 1 79pp. 
1997. An overview of venomous snake evolution. In Thorpe, R.S., Wiister, W. & 

Malhotra, A. (eds) Venomous snakes: ecology, evolution and snakebite (Symposia of 

the Zoological Society of London, No. 70) Pp. 1-13. 
Vaeth, R. H. 1982. Variations in the vidian canal system in Thrasops jacksonii 

(Serpentes, Colubridae). Journal of the Herpetological Association of Africa. 28: 

10-13. 
Villiers, A. 1966. Contribution a la faune du Congo (Brazzaville). Mission A. Villiers 

et A. Descarpentries. XLII. Reptiles Ophidiens. Bulletin de Vlnslitut Francois de 

VAfrique noire, Ser. A, 28: 1720-1760. 
Weinstein, S. A. & Kardong, K. V. 1994. Properties of Duvernoy's secretions from 

opisthoglyphous and aglyphous colubrid snakes. Toxicon 32: 1161-1185. 
Witte, G.-F. de 1953. Reptiles. Exploration du Pare National de I'Upemba, Mission 

G.-F. de Witte, 6: 1-322. 
Wright, D.L., Kardong, K.V. & Bentley, D.L. 1979. The functional anatomy of the 

teeth of the Western Terrestrial Garter Snake, Thamnophis elegans. Herpetologica 

35: 223-233. 

Appendix 1. Material for which skulls were prepared: 

Dispholidus typus (NMZB-UM 3058; NMZB 922, 1350, 1658, 
3322, 10870, 13378); Rhamnophis aethiopissa (NMZB 10793, 
16726); Rhamnophis batesii (NMZB 13206); Xyelodontophis 
uluguruensis (NMZB 7443 - holotype); Thelotomis capensis 
(NMZB-UM 88, 16199. 17922); Thelotomis kirtlandii (NMZB 
32185); Thelotomis mossambicanus (NMZB-UM 3058; NMZB 
11390); Thelotomis usambaricus (NMZB 15629); Thrasops 
flavigularis (NMZB 16725); Thrasops jacksonii (NMZB 10717). 



Appendix 2. Material examined internally: 

Dispholidus typus (BMNH 1979.205; FMNH 58379; IRSL 2 un- 
numbered; IRSNB 13281a-b, MCZ 18223, 32475, 32478, 53458, 
53730-31, 55250, 55255, 67927), Dispholidus 'pemba' (MCZ 
45587), Hapsidophrys lineatus (BMNH 1979.165-67; IRSL 15 
unnumbered; SDSU unnumbered; UNAZA 4 unnumbered; VW 
1010), Hapsidophrys smaragdinus (BMNH 1979.157-59; FMNH 
179036; IRSL 14 unnumbered; MZUSP 8159; PEM 3363, 3403; 
UNAZA 4 unnumbered; VW 907, 1012, 1026, 1068, 1099), 
Philothamnus angolensis (IRSL 23 unnumbered; MZUSP 8 1 74-75, 
8 177; NMV D55548; PEM 3382-83; SDSNH 63865-66; UF52485, 
80395-99, 80671; UNAZA 4 unnumbered; VW 1086, 1197, 1211, 
1254, 1429-30, 1451, 1745, 1986-87, 1990, 2226; ZRC 2.3427), 
Philothamnus bequaerti (MCZ 47846), Philothamnus carinatus 
(UNAZA 2 unnumbered), Philothamnus dorsalis (UNAZA 2 un- 
numbered). Philothamnus heterodermus (UNAZA 1 unnumbered), 
Philothamnus heterolepidotus (LSUMZ 40781), Philothamnus 
hoplogaster (BYU 30895). Philothamnus macrops (MCZ 23244), 
Philothamnus nitidus (UNAZA 1 unnumbered), Philothamnus 
occidentalis (VW 6360), Philothamnus punctatus (MCZ 52666), 
Xyelodontophis uluguruensis (NMZB 7443; ZMB 48153), 
Rhamnophis aethiopissa (IRSL 6 unnumbered; MCZ 1 3607, 258900. 
38392, 48343, 178494; SDSNH 63873; SDSU unnumbered), 
Rhamnophis batesii (IRSNB 28 1 3; MCZ 1 3604, 38393), Thelotomis 
capensis (FMNH 191 163; MCZ41963, 44581, 69036;ZMB 23526), 
Thelotomis kirtlandii (FMNH 205972, 214828; IRSL 1 unnum- 
bered; IRSNB 5370, 5371a-b, 6451, 6454; MCZ 22523, 49687. 
49734, 51835; SDSU unnumbered; ZMB 21627), Thelotomis 
mossambicanus (FMNH 248040; MCZ 51628, 56922; ZMB 16783, 
2800 1 ), Thelotomis usambaricus (MCZ 23349; ZMB 1 6786, 2 1 1 30), 
Thrasops flavigularis (MCZ 8776-77: MHNG 967.20, 1520.68, 
1520.75, 1520.78), Thrasops jacksonii (BMNH 1979.190-91; UF 
52476; MCZ 25954; MZUSP 8178-79; UNAZA 5 unnumbered; 
VW 1077, 1083, 1230, 1232, 1965, 2350), Thrasops occidentalis 
(MCZ 55232; UG C34P12). Thrasops schmidti (MNHN 1940.197, 
1974.1; NRM 2297b). 



Bull. not. Hist. Mus. Land. (Zool.) 68(2): 75-81 



XXC35699M 



Issued 28 November 2002 



On the African leopard whip snake, 
Psammophis leopardinus Bocage, 1887 
(Serpentes, Colubridae), with the description 
of a new species from Zambia 



BARRY HUGHES 

57 Snaresbrook Road, London Ell 1PQ, England. 

E. WADE 

Middlesex University, Cat Hill, Barnet, Hertfordshire. EN4 8HT, England. 



SYNOPSIS. An examination of scalation and dentition of specimens in Brussels (IRSN). Tervuren (MRAC) - mostly Bredo 
collection, and London (BMNH) from Angola. Congo-Kinshasa and Zambia suggests the existence of a species which is neither 
P. sibilans leopardinus of which the type is from Namibia, nor P. 'sibilans' [mossambicus] of Congo-Kinshasa and Zambia, but 
a new species previously unnamed. 



INTRODUCTION 



Bocage (1887:206) described from Catumbela, Angola a Psam- 
mophis (MBL 1798, now destroyed) with a striking reticular pattern 
on the neck and anterior part of the body as a variety of Psammophis 
sibilans, a taxonomic treatment later followed by Broadley (1977). 
More recently Brandstatter (1995, 1996: Fig. 4) has recognised P. 
sibilans as occurring no further south than the northern part of 
Tanzania and has treated Bocage's variety as P. brevirostris 
leopardinus, following an earlier practice by Broadley ( 1 97 1 ). He has 
followed Broadley (op. cit.) in assigning to this subspecies Zambian 
specimens showing the same reticular pattern on the neck. However, 
such a pattern occurs sporadically elsewhere, as in West African 
specimens of P. sibilans (BMNH 1930.6.5.8 from Mogonori, Ghana; 
1956.1.5.87 from Ikoyi, Lagos, Nigeria; CM 24636 from Accra; 
MNHN 1 985.442-3 from Ghana; ZMH R04466 from Gana Gana or 
Segbana, Niger Delta, Nigeria: these have neck bars sometimes 
interconnected as in leopardinus. Dependence on pattern for identifi- 
cation in a genus whose species are notorious for their variability is 
unconvincing. In an attempt to find other, more reliable criteria by 
which to distinguish species of Psammophis, total tooth counts were 
undertaken and revealed significant differences between specimens 
of ''leopardinus'' from Angola and those from Zambia. Secondly, the 
Zambian specimens are often of a colour pattern rarely met with 
elsewhere during the study of several thousand specimens from all 
parts of Africa and the Middle East. Thirdly, the ventral and subcaudal 
counts of the Zambian specimens are lower than those from neigh- 
bouring localities in Zambia and Congo-Kinshasa. Fourthly, a SEM 
micrograph of a dorsal scale of a specimen from Ikelenge (Brandstatter, 
1995: Fig. 39) differs considerably from those of species assigned to 
the P. sibilans complex. For these reasons, it is thought necessary to 
coin a new name for the Zambian specimens. 



SYSTEMATICS 



Psammophis zambiensis sp. nov. Zambian Whip Snake 
Psammophis sibilans, not Linnaeus 1758, Pitman, 1934: 297 (part. 



Chimikombe specimens only). 
© The Natural History Museum, 2002 



Dromophis lineatus, not Dumeril & Bibron, Laurent 1956:247 
Kundelungu male & female. 

Psammophis ? sibilans Broadley & Pitman 1960: 445 

Psammophis brevirostris leopardinus Broadley 1971:88; 
Brandstatter, 1995: 53, Fig. 39 and 1996: 48 (Zambian specimens 
only); Haagner el al. 2000: 16. 

Psammophis sibilans leopardinus (Zambian specimens) Broadley 
1977:18 

Psammophis brevirostris leopardinus Brandstatter 1996:48 (Zam- 
bian specimens only) 

Holotype. BMNH 1 959. 1.1.81 supposedly from ' Abercorn' (now 
Mbala) area of Zambia, part of the H.J. Bredo collection, sent on 
from Brussels, but likely to be from Mweru-Wantipa - see discussion 
(Figs 1-3). 

PARATYPES. IRSN 2561,2565-6 of same origin, BMNH 
1932.9.9.132-3 from Chimikombe at 4500 ft. (= Chimilombe, 
Solwezi District); NMZB 10635-6, 10736, 10757 from Ikelenge 
(Broadley 1991:529); IRSN 2562 from Mambwe; IRSN 2567 and 
PEM 1438/ 12 from Mporokoso District (probably Mweru-Wantipa); 
IRSN 2563 from Mweru-Wantipa, and IRSN 2564 from an unknown 
source in Zambia; MRAC 18622-3 SERAM, Kundelungu Plateau 
1750 m, Congo-Kinshasa (Laurent 1956:247 as Dromophis lineatus). 
All specimens, except two (BMNH 1932.9.9.132-3) are female; 
Haagner et al (2000) have listed two more males as i P brevirostris 
leopardinus'. 

Diagnosis. Often distinguished by a combination of the reticular 
body pattern of leopardinus but lacking the higher tooth counts of 
the latter (Table L). A detailed description of colouration, based on 
5 specimens, is given by Broadley & Pitman (1960:445) but can be 
summed up by saying that they are greenish rather than the usual 
khaki-brown and the scales heavily edged in black. Unlike associ- 
ated specimens of P.'sibilans' the vertebral 'chain' is more like a 
stripe, the lighter marking on each vertebral scale being more of a 
line than a spot; and behind the eyes the head is crossed by three 
transverse light bars - a common feature in many Psammophis spp. 
but these are narrow, as in Pangolensis or Dromophis lineatus. 
Smaller specimens (e.g. Fig. 2-3) are more distinctly marked with 
greater contrast around the body. As Haagner et al (2000) have 



76 



B. HUGHES AND E. WADE 






Fig. 1 Head of P. zambiensis (adult holotype, BMNH 1959. 1 . 1 .8 1 ) seen in (a) dorsal, (b) lateral and (c) ventral views. 



Table 1 Dentitions - left/right sides. 



species 



museum no. 



max.pre-2F 



post.dentary 



palatine 



pterygoid 



leopardinus 

(Namibia) 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

zambiensis 

sibilans 



1937.12.3.166 

1959.1.1.81 

10636 

10736 

10521 

10522 

10523 

18622 

18623 

1932.9.9.132 

1932.9.9.133 

1953.1.2.15 

1953.1.2.14 



5/5 

4/3 
-/3 
4/4 
3/3 
3/3 

3/3 
4/4 
3/3 
3/3 
3/3 
3/3 



23/24 

17/14 
14/13 
15/14 
15/15 
13/- 

16/16 
17/15 
17/19 

20/20 
17/18 
19/18 



11/9 

9/8 
8/9 
8/9 
8/8 
9/8 

7/8 
8/9 
9/9 
8/8 
8/8 
8/8 



23/18 

17/14 
13/12 
16/15 
16/16 
15/14 

14/15 
17/18 
17/16 
19/18 
18/18 
16/15? 



A NEW PSAMMOPHIS FROM ZAMBIA 



77 






Fig. 2 Head ot'juvenile paratype P.zambiensis (IRSN 10523) seen in (a) dorsal, (b) lateral, and (c) ventral views. 



noticed, the reticular neck pattern is not always present and these 
specimens are distinguished from 'sibilans' by their lower ventral 
counts and usually by lower subcaudal counts (Table 2). 

Brandstatter (1995: Fig. 39) has provided a SEM micrograph of a 
dorsal scale from a P. zambiensis paratype NMZB 10636, and the 
micro-ornamentation resembles that of Dromophis lineatus (his Fig. 
83) more than any species of the P. sibilans complex. 

HABITAT. Unfortunately, no field notes are available for this species, 
but the fact that many specimens appear to have originated from the 
Mweru-Wantipa suggests that it requires a marshy habitat like 
Dromophis lineatus, with which it is sympatric in this area (Broadley 
& Pitman, 1959). In the Ikelenge area there there are many suitable 
dambos and one local specimen had eaten an Eumecia anchietae, a 
large skink that frequents such places (Broadley, 1991). The 



Sanolumba snake had eaten a ranid frog (Haagner et ai, 2000). 

Other species and sources of data 

Psammophis leopardinus [only those with numbers seen by BH, 
those without numbers DGB data or from publications]. 

ANGOLA - Bella Vista MCZ; Caconda MBL x 8; Capelongo 
AMNH x 6 ; Catengue SMF x 2; Catumbela MBL lectotype, 
destroyed; Iona TM; Luanda USNM; Lobito Bay AMNH R50612- 
3, R506 1 7-8, and x5; Oncocua, 37 km NE on way to Otchinzau TM; 
NAMIBIA - Swakop-Tal, Namib Desert BMNH 1937.12.3.166. 

Psammophis 'sibilans', currently treated as P. mossambicus. 

CONGO-KINSHASA - Kambore MRAC 2017; Kansenia MRAC 
7002, 7639; Kapanza MRAC 9649-50; Kapiri MRAC 7027, 7056- 



78 



B. HUGHES AND E. WADE 





Fig. 3 Stretches of the body of P.zambiensis between ventral scales 50 and 57 seen in dorsal and ventral views, (a, b) Adult holotype (BMNH 
1959.1.1.81); (c,d) juvenile paratype (IRSN 10523). 



A NEW PSAMMOPHIS FROM ZAMBIA 
Table 2 Scale counts (sample size) 



species 



M ventrals 



F ventrals 



M subc. 



F subc. 



'leopardinus' 
Heop.'' refined 
zambiensis 
'sibilans' 



151-71 (9) 
151-65(8) 
148-61 (5) 



151-74(20) 
151-67(16) 
149-65(17) 



167-77(32) 167-77(19) 



79-104(4) 80-105 (10) 

79-104(4) 80-105 (8) 

80-90 (3) 75-86 (9) 

89-103(26) 781-100(12) 



N.B. The P. 'leopardinus' data is for Angolan specimens and from Broadley (pers. 
com.): I suspect that specimens of another species are included and P. 'leop. refined' 
has the data of that species removed. The P. 'sibilans' (currently treated as P. 
mossambicus) data is from Zambian specimens so called by Broadley (1971:88) 
although he has since referred them to P. phillipsi (Broadley 1983) and later P. 
mossambicus (Broadley. in prep.), and from Haagner et al (2000) who treat their 
specimens as P. mossambicus. The P. zambiensis data incorporates that of P. 
'breviroslris leopardinus' from Haagner et al, (2000). 




Fig. 4 Brandstiitter's (1996, fig. at p. 48) map of the occurrence of 
Psammophis breviroslris leopardinus. 



79 

BMNH 97.6.9.131 (Boulenger 1897:801); Zomba BMNH 
93.10.26.57, 94.2.13.12 (Gunther 1894:618. Boulenger 
1896:164,l,m), 1933.4.5.2; Zomba Mt BMNH 1948.1.2.28. 

NAMIBIA - Old Sangwali (Broadley 1983; Barts & Haacke 
1997). 

TANZANIA - Ipiama ZMB 16984; Kingani, nr Dunda ZMB 
172777, 17338; Zimba ZMB 23476; 

ZAMBIA - IRSN 8834,b-c, 10520, Broadley & Pitman 1960; 
ZFMK 18904; Barotseland MNHN 1921.533; Buleya IRSN 8802 
(Bulaya of Broadley & Pitman 1960); Chipangali UM; Chisi Lake 
(Broadley & Pitman 1960); Chunga, Kafue N.P UM; Dumdumwensi 
UM 20841 ; Fort Manning BMNH 1 962.497; Ikelenge x 2 (Broadley 
1991 as Rphillipsii); Kabinda BMNH 1932.9.9.130; Kabwe (as 
Broken Hill) BMNH 1932.5.3.95-100, 1932.9.9.134-8, 1936.3.6.34, 
1959.1.1.96; Kalabo FMNH, UM; Kaputa IRSN 8805,a, Broadley 
& Pitman 1960); Kasama IRSN 8832, Broadley & Pitman 1960; 
Lachisi IRSN 8830; Lealui MNHN 20. 104, 2 1 ;533; Lusaka, 100 km 
SW of ZFMK 18904; Makupa IRSN 8835,a,b; Mambwe IRSN 
8804, 8828-9, Broadley & Pitman 1960; Maskie's, Namwala Dis- 
trict BMNH 1932.5.3.95-8. 1932.5.3.101; Mbala (as Abercorn, 
Broadley & Pitman 1960) BMNH 1959.1.1.81, 1959.1.1.96; IRSN 
8798a-e - 803,a, 8799a-b, 8800-1, 8806-27. 8836-8, 8839 
(Broadley & Pitman 1 960 as Psammophis subtaeniatus sudanensis); 
Mkanda UM; Mporosoko IRSN 8803a-b (Mporokoso of Broadley 
& Pitman I960); Msoro UM (Wilson 1965); Mukupa (Broadley & 
Pitman 1960); Muswema IRSN 8833. Broadley & Pitman 1960; 
Mweru-Wantipa IRSN 8831, Broadley & Pitman 1960; 
Namantombwa Hill, Mumbwa UNZA; Nchelenge, Luangwa Valley 
BMNH 1932.12.13.231; Ngoma. Kafue N.P. UM: Nsangu BMNH 
1932.9.9.131; Sayiri UM (Wilson 1965); Serenje BMNH 
1953.1.2.13-6 ; Yacobi Village. Luangwa Valley LACM; 

ZIMBABWE - Bari. Chikwakwa UM; Bulawayo UM x 3; Elim 
Mission, Inyanga UM; Harare (as Salisbury) NMK; Harare (as 
Salisbury). Borrowdale Brook UM x 2; Inkomo UM; Mabalauta 
UM; Mazoe (Broadley 1959) BMNH 1902.2.12.96-7; Mondoro 
UM; Odzi UM x 2; Rugare, Inyanga UM; Umsweswe Bridge, 
Gatooma UM; Umtali UM x 3 (Broadley 1959) BMNH 1954. 1 .3.23- 
4; Vumba Mt UM x 2: Wankie N.P, main camp UM x 2. 

Psammophis zambiensis (other, non-type specimens) 

ZAMBIA - Ikelenge NMZB 10636, and x 2; PEM x 7; Mbala (as 
Abercorn, Broadley & Pitman 1960) IRSN 10521-2; Mbala area (as 
'Abercorn' but see Discussion); BMNH 1959.1.1.81; Mporosoko 
(Broadley & Pitman 1960 as Mporokoso) IRSN 10523; PEM x 2 
( Haagner et al 200): Sakeji School PEM x 6 (Haagner et al., 2000). 



9, 7061; Kapolowe MRAC 9970;Kasai MRAC 968; Kasenyi IRSN 
6861; Lofoi MRAC 598; Lubumbashi [as Elizabethville] IRSN 
63 10; MRAC 766 1 , 8378-9, 9397-8; Lukafu MRAC 7 1 87, 7 1 199- 
201, 7217; Luluaborg St Joseph MRAC 2627-9; Luebo MRAC 
2996; Lukonzolwa MRAC 2165; Lusambo MRAC 16378, 16381; 
Merode MRAC 3113; Moero Lake Region MRAC 15323; Musosa 
IRSN 4780-1 ; Niambi to Baudouinville MRAC; Pweto MRAC 252, 
260, 1980, 1999, 2027; Sandoa MRAC 7935, 7941, 7967-9, 8271- 
2, 9854; Tembwe MRAC 4186, 4216, 4237-8; 

MALAWI - Chitipa (as Fort Hill) BMNH 97.6.9. 1 35-6; Chiromo 
BMNH 1959.1.3.37; Chiromo, 20 km N of Mangochi (as Fort 
Johnston); Fort Johnston BMNH 1926.5.8.49; Kasungu AMNH; 
Kondowe (=Livingstonia Mission) to Karonga BMNH 97.6.9.132- 
4; Nkhotakhota (as Kota Kota) BMNH 96. 12.12.19; Mkanga BMNH 
1959.1.3.38; Mlanje River AMNH; x 2; Mtimbuka AMNH x 4; 
Mulanje Mt. AMNH; Nchisi AMNH x 2; Nkahta Bay to Ruarwe 



DISCUSSION 



The many names by which Zambian P.'sibilans' has been known 
(see above under synonymy) is an indication of the uncertainly 
which attends identification of specimens of this species complex. 
The very distinctive colouration of some specimens of P. zambiensis 
attracts attention but it is not reliable in separating this species from 
other(s) with which it may be sympatric. A letter from Desmond 
Vesey-FitzGerald to Donald G. Broadley (Broadley, pers. comm.), 
dated 29 Sept. 1959, suggests that the source of 'Abercorn' speci- 
mens is to be doubted: 'I would guess that all these snakes may have 
come from Mweru-Wantipa in Mporokoso District, where Bredo 
would have been collecting in the 1943/44 period.' Vesey-FitzGerald 
(1958) collected long series of P. sibilans [= P. mossambicus] in 



80 



B. HUGHES AND E. WADE 



1 

20 

/ 
.') 

-10°-- 

-15° — 
-20°- 


I I I I I 

2; 

• 

• 
/ 

XT"" "\ 

/ CONGO 

\ 
i 

! 

i 
i 

\ • 


1 1 1 I \ I i I I I I I 

;° ) i 30° : 
-KINSHASA I A ^\ ^ 

•• • H& ,v ^ * 

in \ o A 


l l 
!5° 

*• 


10° 


i 
ANGOLA \ 

i 

i 

▲ 






P 

O IS-- 


• • 

\ 
\ 

\ ° 
\. jQ._. 

-_ ::" °o,c 

b~"°~ v 

B T S W A 


• o 

• • c 
o oo 
o • 

J 
OO 0°Q cPfCpf O 

o rOrJ 

iX'-a-./ o 
■\ 

\ Z I M B A B 
\ 

N A 


1 ft ft 

°88$3 / * * ^< 

W E / J^ 

500 km 

i i i i i i 


1 "\._ 


i > l' w/ 


on° - 


20° i 25° \ 30° 7 35° 

1 i i i I I 1 l \ 1 1 i 1 I I I /I 1 1 1 



Fig. 5 Map of area of sympatry between P.'sibilans' (O, literature ref. which may include zambiensis; • specimen seen by BH or DGB and P.zambiensis 
▲ specimens seen). Many records taken from Broadley (1983:146. map 36 of P. 'phillipsii', others from an as yet unpublished, revised map (DGB in 
prep.) which we have been privileged to see. Localities listed by country and quarter degree square (without 'se' prefix); sources indicated when locality 
is a map plot without name. For locality data see Appendix 1. 



Abercorn [= Mbala] District, but none had either the characteristic 
pattern or low ventral counts of P. zambiensis. However, he did 
record one snake from Chinsali (10.32 CI) with only 160 ventrals 
and 90 subcaudals, which may have been a P. zambiensis, but it was 
apparently not preserved. Only by collecting data on a large number 
of specimens can the limits of variability become known and 
consistent differences in meristic data become apparent. 

P. zambiensis and P. i sibilans'' appear to be sympatric at Mbala 
(Abercorn) (Fig. 3, se08.31C4) but all of the P. zambiensis speci- 
mens so attributed are likely from Mweru-Wantipa (see above), so 
that true sympatry may occur only at Mporokoso (se09.30), Ikelenge 
(sell.24a2) Serenje (sel3.3061) and near Mchinji (13.32d4). The 
co-occurrence over such a large distance - more than 600 km from 
Ikelenge to Mweru Wantipa without more instances of sympatry 
suggests the occupation of different habitats. 

P. zambiensis seems to be distinct from P. leopardinus to the south 
and P.' sibilans'' (or i phillipsV) to the north; the true identity of the 



latter can become clearer only after analysis of specimens from the 
whole of the Congo Basin and West Africa. 

ACKNOWLEDGEMENTS. We are indebted to Donald Broadley for his usual 
generosity with his data and advice and for the loan of specimens; to curators 
Charles Myers and Richard Zweifel (AMNH), Georges Lenglet (IRSN, 
Brussels). Danny Meirte (MRAC. Tervuren). Rainer Giinther (ZMB), and 
Colin McCarthy (BMNH) for similar loans and providing one or both of us 
with working space and answering our many queries. 

REFERENCES 

Barts, M & W.D. Haacke. 1997. Zur Reptilienfauna der Tsodilo-Berge und 

Angrenzender Gebiete im NW-Botswana. Teil 2: Sauria: 2 Serpentes. Sauria 19:15- 

21. 
Bocage, J.V.B. du. 1 887. Melanges erpetologiques. IV. Reptiles du dernier voyage de 

MM. Capello et Ivens a travers l'Afrique. Jornal de Sciencias Mathematicas, 

Physicas e naturaes. Lisboa. 10: 201-208. 



A NEW PSAMM0PH1S FROM ZAMBIA 



81 



Boulenger, G.A. 1 896. Catalogue of the snakes in the British Museum (Natural 
History). 3, 727 p. 

1 897. A list of the reptiles and amphibians collected in Northern Nyasaland by 

Mr.Alex Whyte. F.Z.S. and presented to the British Museum by Sir Henry H. 
Johnston, K.C.B., with descriptions of new species. Proceedings of the Zoological 
Society of London [1897]: 800-803. 

1910. A revised list of the South African reptiles and batrachians, with synoptic 

tables, special reference to the specimens in the South African Museum and descrip- 
tions of new species. Annals of the South African Museum. 5: 455-538. 

Brandstatter, Frank. 1995. Eine Revision tier Gaining Psammophis mil 
Berucksichtigung der Schwesterngattungen innerhalf der Tribus Psammophiini 
(Colubridae; Lycodontinae). Dissertation zur Erlangung des Grades des Doktors der 
Naturwissenschaften der Mathematisch-Naturwissenschaftlichen Fakultiit der 
Universitiit des Saarlandes. Saarbriicken. 480 pp. 

1996. Die Sandrennattern. Die Neue Brehm-Biichkerei 636.. Westarp 

Wissenschaften, Magdeburg. 142 pp. 

Broadley, D.G. 1959. The herpetology of Southern Rhodesia Part I. Snakes. Bulletin 

of the Museum of Comparative Zoology 120: I — 1 00. 

1971. The reptiles and amphibians of Zambia. Puku (6): 1-143. 

1977. A review of the Genus Psammophis in southern Africa (Serpentes: 

Colubridae). Amoldia Rhodesia 8(12): 1-29. 

1983, 1990. FilzSimons' snakes of southern Africa. 387 p. Jonathan Ball & Ad. 

Donker, Parklands. South Africa. 

1991. The herpetofauna of northern Mwinilunga District. Northwestern Zambia. 

Amoldia Zimbabwe 9 (37): 519-538. 
In prep. A review of the species of Psammophis south of latitude 12 S degrees 

(Serpentes: Psammophiinae). 
& Pitman, C.R.S. 1959. On a collection of snakes taken in Northern Rhodesia by 

Monsieur H.J. Bredo. Occasional Papers of the National Museums of Southern 

Rhodesia (24B): 437-451. 
Gunther, A. 1894. Second report on the reptiles, batrachians, and fishes transmitted by 

Mr. H.H. Johnston. C.B.. from British Central Africa. Proceedings of the Zoological 

Society of London [1893]: 616-628. 
Haagner, G.V., Branch, W.R. & Haagner, A.J.F. 2000. Notes on a collection of 

reptiles from Zambia and adjacent areas of the Democratic Republic of the Congo. 

Annals of the Eastern Cape Museums 1: 1-25. 
Johnsen, P. 1962. Notes on African snakes, mainly from Northern Rhodesia and 

Liberia. Videnskabelige Meddelelserfra DanskNaturhistorisk Forening I Kobenhavn 

124:115-130. 
Laurent, R.F. 1956. Contribution a I'herpetologie de la region des Grands Lacs de 

l'Afrique centrale. Annates du Musee royal du Congo Beige (ser. 8) 48: 1-390. 
Norton, C.C. & Peirce, M.A. 1985. Caryospora species from Zambian snakes. African 

Journal of Ecology 23: 59-62. 
Peirce, M.A. 1984. Some parasites of reptiles from Zambia and Indian Ocean islands 

with a description of Haemogregarina zambiensis sp. nov. from Dispholidus typus 

(Colubridae). Journal of Natural History 18:21 1-217. 
Pickersgill, M. & C.A. Watson. 1998. Report on the reptiles observed on a field trip 

to Eastern Africa, October 1996-August 1997. Herptile 23:145-149. 
Pitman, C.R.S. 1934. A check list of Reptilia and Amphibia occurring and believed to 

occur in Northern Rhodesia. In: A report on a faunal survey of Northern Rhodesia. 

pp. 292-312. Government Printer. Livingstone. 
Sweeney, R.C.H. 1961. Snakes of Nyasaland. Zomba: Nyasaland Society. 200 pp. 
Swynnerton, G.H. 1957. Notes on the fauna. Reptiles. Annual Report oj the Game 

Preservation Department of Tanganyika Territory ( 1955-6): 28-29. 
Vesey-FitzGerald, D.F. 1958. The snakes of Northern Rhodesia and the Tanganyika 

borderlands. Proceeding and Transactions of the Rhodesia Scientific Association. 

46: 17-102. 
Wilson, V.J. 1965. The snakes of the Eastern Province of Zambia. Puku 3: 149-170. 



Appendix 1 

ANGOLA - 17.19d3 DGB in prep. BOTSWANA - 18.21b4 DGB 
in prep.; cl DGB in prep. CONGO KINSHASA - 06.23a4 Merode; 
06.24c3 Kapanza; 06.29c4 Tembwe; 07.28a3 Kiambi; 07,29 Niambi 
to Baudouinville; 07.29b2 Baudouinville; 08.25c3 Kamina; 08.26a2 
Kikondja. 08.26c2 Nyonga; 08.28b4 Pweto; 08.28d3 Lukonzolwa;, 
09.22d3 Sandoa; 09.28b2 Moero Lake region; 10.22a4 Dilolo; 
10.26a3 Kansenia; Kapiri; 10.26d3 Kambove; 10.27 Katanga; 
10.27a2 Lofoi; 10.27dl Lukafu; 10.27d2 SERAM, Kundelungu; 
10.28b3 Kasenga; 11.26b2 Kapolowe; 11.27c2 Lubumbashi [as 
Elizabethville] 11.24bl Sanolumba ; MALAWI - 09.33c2 Chitipa 
(as Fort Hill); 09.33dl Misuku Hills; 09.33d4 Karonga (Pickersgill 



& Watson 1998); 10.34a3 Nyungwe; 10.34cl Kondowe ( = 
Livingstonia Mission); 10.34c3 Nchenachena; 11.33b2 Rumpi; 
11.34a3 Nkhata Bay to Ruarwe; 12.34c4 Nkhotakhota (as Kota 
Kota); 13.32a2 Kasungu; 13.34a3 Nchisi Mt.; 14.34a4 Dedza; 
14.35a3 Mtimbuka; a4 Mangochi (as Fort Johnston); 15.35a4 
Zomba; b3 Mchenga; c3 Limbe (Sweeney 1961:147); d3 Mlanje 
Mt.; 16.34b2 Likabula R., Mt Mlanje; 16.35a3 Broadley (1983, as 
P.phillipsii); bl Broadley (1983 as P.phillipsii); cl Chiromo 
(Broadley 1983 as P.phillipsii); Makanga. MOZAMBIQUE - 
16.31bl DGB in prep; 17.35d3 Broadley (1983); 18.33a3 Broadley 
(1983); 18.34a4 Broadley (1983); 18.35b4 Broadley (1983); NA- 
MIBIA - 17.24cl DGB in prep.; c2 DGB in prep.; c4 DGB in prep.; 
d4 DGB in prep.; 17.25c3 DGB in prep.; 18.23b3 Old Sangwali; 
TANZANIA - 07.31d4 Zimba: 08.31b2 Milepa; 08.35b4 Uzungwa 
(as Udzungwa Mts); 09.33d2 Ipiama. ZAMBIA - 08.29c4 Mukupa 
Katandula; 08.29d3 Lake Chisi, Mweru-Wantipa; 08.29d4 Kaputa; 
08.31c4 Mbala (as Abercorn); 09.28b2 Molo; 09.29c3 Kawambwa; 
09.30a3 Mporokoso; 09.31b2 Mambwe; 10.30c3 Luwingu District: 
10.31a2 Kasama; 10.31bl Nchelenge; 11.24a2 Ikelenge; Sakeji 
School; 11.31c4 Mpika; Nsangu R.; 12.26b3 Chimilombe (as 
Chimikombe); 12.27b4 Chingola (Haagner et ah, 2000); dl 
Lufwanyama Farm (Haagner etal., 2000); d2 Musenga (Haagner et 
al., 2000); 12.28c3 Kitwe (Haagner et al, 2000): d3 Ndola (Johnsen, 
1962); 12.28cl 24 km W of Mufulira (Johnsen, 1962); 12.30a2 
Kabinda. Lukulu R.; 12.32cl Yakobe (as Yacobi), Luangwa Valley; 
12.33a3 Lundaz (Wilson. 1965)i; 12.34c4 Kota Kola; 13.24cl 
Kabompo.; 13.25b4 Kasempa; 13.28a2 8km W. of Luanshya.; 
13.30a2 DGB in prep.; a3 Katete (Wilson, 1965); bl Serenje - P. 
zambiensis sympatric with P. ' sibilans' '; 13.31d2 Msoro; b3 Nsangu 
River, W of Lavushi River; 13.32a3 Chikowa (Wilson, 1965); bl 
Chipangali (Wilson, 1 965); b3 DGB in prep.; c2 Kalichero (Wilson, 
1965); c4 Sayiri (Wilson, 1965); dl Chipata (Wilson 1965, as Fort 
Jameson); d2 Mkanda; d4 near Mchinji (as Fort Manning): 13.33a2 
Kasanga; 14.22d3 Kalabo; 14.27c3 Mumbwa; 14.28a4 Kabwe (as 
Broken Hill);14.30dl Kacholola; 15.22bl Buleya; b4Ndau School; 
15.23al Lealui; 15.25d4 Ngoma, Kafue N.P.; 15.26al Chunga, 
Kafue N.P; c2 Maskies; Fort Manning (Wilson, 1965); 15.27d3 
100 km SW of Lusaka; d4 Mazabuka.; 15.28a4 Lusaka; b3 50 km 
E. of Lusaka; c2 Balmoral, (Peirce. 1984; Norton & Peirce, 1985) 
Chilanga; 15.29c2 DGB in prep.; d2 DGB in prep.; 16.23dl Kachola 
(Wilson 1965, as Kacholola); 16.25c3 Machile Forest Station.; c4 
Mulanga;d4 Katanda; 16.26c I Dumdumwensi;c4Kasusu; 16.27c3 
Nansi Farm; d4 Chezia confluence, Kariba Lake; 17.24a3 Katima 
Mulilo; a4 Wenela base, Caprivi; 17.26a2 Kalomo; c4 DGB in 
prep.; d4 Ihaha; 17.27bl Kariba lake, Lulongwe confluence; 
18.21b4 Broadley (1983); cl Broadley (1983); 18.23b3 Old 
Sangwali; ZIMBABWE - 16.28c4 Lake Kariba, Bumi confluence; 
dl Kariba lake, Sanyati basin; 16.29d3 DGB in prep; 16.30b4 
Mzarambanitam; c3 Salator Farm, Mkanguru; 16.31d3 Mt Darwin 
(Broadley, 1959); 17.28c3 Sengwa R.; 17.30al DGB in prep.; a3 
Sinoia (Broadley, 1959); bl Msitkwe R., Mutorashanga; cl 
Selukwe; c2 Broadley (1983); c3 Kutama (Broadley, 1959);c4 
Broadley (1983); dl Inkomo, d2 Mazoe (Boulenger 1910; 
(Broadley, 1959); Mt Hampden (Broadley, 1959); d3 Norton 
(Broadley, 1959); d4 Hunyani (Broadley, 1959); 17.31a3 Broadley 
(1983); cl Harare (as Salisbury, Broadley 1959); Borrowdale Brook; 
c2 Bari, Chikwakwa; c3 Harare (as Salisbury, Boulenger 1910); c4 
Melfort (Broadley, 1983); 17.32a3 5 km W of Mutolo Broadley 
(1983); d2 Elim Mission, Inyanga; d3 Maristvale, Nyanga (Broadley 
1983); d4 Nyamaropa, Nyanga (Broadley, 1983); 18.26c4 Broadley 
(1983); d2 Broadley (1983); 18.27c2 Broadley (1983); 18.32b4 
Broadley (1983). 



Bull. mil. Hist. Mus. Loud. (Zool.) 68(2): 83-90 



Issued 28 November 2002 



Morphological variation and the definition of 
species in the snake genus Tropidophis 
(Serpentes, Tropidophiidae) 

S. BLAIR HEDGES 

Department of Biology, 208 Mueller Laboratory, The Pennsylvania State University, University Park. 
Pennsylvania 16802, USA. e-mail sbhl@psu.edu 

SYNOPSIS. Historically, the definition of species in the Neotropical snake genus Tropidophis has been difficult because of 
intraspecific variation in scalation and a paucity of specimens of most taxa. There were 13 species recognized at the time of the 
last review in 1960. but additional species have since been discovered and a taxonomic review and update is needed. Data on 
morphological variation are presented here and used to clarify the status of the described taxa. Because many taxa are allopatric 
with their closest relatives, it is necessary to make decisions as to their status as species or subspecies. As a gauge of species status 
in the genus, character divergence in ten pairs of closely related sympatric species was examined. Typically, such species are 
differentiated by two non-overlapping colour pattern differences, often in combination with a diagnostic (non-overlapping) or 
overlapping difference in scalation. Using this criterion, seven taxa previously considered as subspecies are here elevated to 
species status, whereas seven other taxa are retained as subspecies, although in some cases they are allocated to different species. 
As a result, the genus Tropidophis is considered here to comprise 29 species, 26 of which are West Indian and 15 of those are 
restricted to Cuba. 



INTRODUCTION 



Tropidophis are typically small, stout-bodied snakes of the family 
Tropidophiidae that occur in South America and the West Indies. 
This family is a member of the primitive snake Infraorder Henophidia 
(Underwood, 1967). As recognized here, there are 29 valid species 
oi'Tropidophis and all but three occur in the West Indies, where Cuba 
(15 species) is the centre of diversity (Table 1). They are nocturnal 
and feed mostly on sleeping lizards (especially Anolis), but also on 
frogs (especially Eleutherodactylus); other nocturnal snakes may 
impinge on Tropidophis ecologically. All are viviparous and most 
are terrestrial, although several Cuban species are arboreal and 
gracile in habitus. They exhibit a diversity of colour patterns that 
include spots (mostly), bands (saddles), and stripes. They have the 
unusual ability of being able to change their colouration, physiologi- 
cally (Hedges. Hass & Maugel, 1 989). Typically they are paler when 
active (at night) and dark while inactive. Species distributions tend 
to be greatly restricted, with species endemic to single islands or 
island banks, and often to small areas on an island. However, species 
density can be high, and as many as six species are sympatric in 
some areas of Cuba. 

Historically, the taxonomy of Tropidophis has been difficult to 
study because of small numbers of specimens and a paucity of 
diagnostic characters. For example, two of the earliest described 
species, T maculatus and T pardalis, have been confused repeat- 
edly. Boulenger ( 1 893) and Stull ( 1 928) commented on the confusion 
of these species by Cope (1868), whereas Schwartz and Marsh 
( 1 960) later commented on their confusion by Stull ! Most of these 
early problems in Tropidophis taxonomy stemmed from the use of 
characters later found to be unreliable, such as the keeling of scales 
or hemipene morphology. It was not until Schwartz and Marsh 
( 1 960) assembled a large number of specimens and collected exten- 
sive data on proportions, scalation and pattern that the systematics of 
this genus became reasonably well known. Although it was a large 
study, it was not comprehensive because it omitted species related to 



T melanurus and those placed by Schwartz (1957) in the semicinctis 
group. However, their success was in recognizing the utility of 
colouration and pattern characters, and that species diagnosis in this 
genus often requires consideration of multiple characters, some of 
which may not be individually diagnostic. 

This is not a comprehensive revision of the genus but rather a 
taxonomic update, motivated by the many changes that have occurred 
since that last major review (Schwartz & Marsh, 1960) and the need 
to summarize what is known of morphological variation in the 
genus. Another motivation is to address a recurring problem in the 
systematics of this group: determining the species status of allopatric 
populations and taxa. In the process, taxa previously considered as 
subspecies are here elevated to species status, some are assigned to 
different species, and others are left unchanged. 



MATERIALS AND METHODS 

The data presented herein are almost entirely from the literature, or 
were used in published studies (but not necessarily published in the 
form here). Most derive from the raw data sheets of the late Albert 
Schwartz, used primarily in several publications (Schwartz, 1975; 
Schwartz & Garrido, 1975; Schwartz & Henderson, 1991; Schwartz 
& Marsh, 1960; Schwartz & Thomas, 1960; Thomas, 1963). 
Schwartz's Cuban specimens are in the American Museum of 
Natural History and his other material is almost entirely in the 
collection of the Museum of Natural History, University of Kansas. 
In addition to those data, I have included data from specimens I and 
colleagues have collected during the last two decades of field work, 
and which, for the most part, formed the basis of several published 
studies: (Hedges, Estrada & Diaz, 1999; Hedges & Garrido, 1992; 
Hedges & Garrido, 1999; Hedges & Garrido, 2002; Hedges, Garrido 
& Diaz, 2001). This material is in the National Museum of Natural 
History (Smithsonian) and in Cuban collections (National Museum 
of Natural History, Havana; Institute of Ecology and Systematics, 



© The Natural History Museum. 2002 



84 



S.B. HEDGES 



Table 1 Species, species groups, and distributions of snakes of the genus 
Tropidophis. 



Species 


Species Group 


Distribution 


T. battersbyi Laurent 


taczanowskyi 


South America 


T. bucculentiis Cope 


melanurus 


Navassa Island 


T. canus Cope 


melanurus 


Bahamas 


T. caymanensis Battersby 


melanurus 


Grand Cayman 


T. celiae Hedges, Estrada, and Diaz 


melanurus 


Cuba 


T. curtus Garman 


melanurus 


Bahamas 


T. feicki Schwartz 


maculatus 


Cuba 


T. fuscus Hedges and Garrido 


pardalis 


Cuba 


T. galacelidus Schwartz and Garrido 


pardalis 


Cuba 


T. greenwayi Barbour and Shreve 


haetianus 


Turks and Caicos 


T. haetianus Cope 


haetianus 


Hispaniola 


T. hardyi Schwartz and Garrido 


pardalis 


Cuba 


T. hendersoni Hedges and Garrido 


pardalis 


Cuba 


T. jamaicensis Stull 


jamaicensis 


Jamaica 


T. maculatus Bibron 


maculatus 


Cuba 


T. melanurus Schlegel 


melanurus 


Cuba 


T. morenoi Hedges, Garrido, and Diaz 


maculatus 


Cuba 


T. nigriventris Bailey 


pardalis 


Cuba 


T. pardalis Gundlach 


pardalis 


Cuba 


T. parked Grant 


melanurus 


Little Cayman 


T. paucisquamis Miiller 


taczanowskyi 


South America 


T. pilsbryi Bailey 


pardalis 


Cuba 


T. schwartzi Thomas 


melanurus 


Cayman Brae 


T. semicinctus Gundlach and Peters 


maculatus 


Cuba 


T. spiritus Hedges and Garrido 


pardalis 


Cuba 


T. stejnegeri Grant 


jamaicensis 


Jamaica 


T. stullae Grant 


jamaicensis 


Jamaica 


T. taczanowskyi Steindachner 


taczanowskyi 


South America 


T. wrighti Stull 


pardalis 


Cuba 



Havana). In nearly all cases, museum numbers and localities of 
those specimens are listed in the publications and therefore are not 
repeated here. 

In some cases, summary data presented in the tables of Schwartz 
and Marsh (1960) do not agree with those in the raw data sheets or 
with data mentioned in the text of Schwartz and Marsh, presumably 
because of typographical errors in their tables. Some of the data 
presented later in Schwartz and Henderson (1991), such as the 
ventral range of T. canus and caudal range of T. maculatus, appear to 
be derived from those typographical errors. Although these errors 
are minor, the summary data presented in this paper were taken 
directly from Schwartz's raw data sheets, to avoid any confusion, 
and supplemented with additional data. Also, some characters were 
not scored by Schwartz in some species (e.g., parietal contact in T. 
feicki, T. melanurus, T. semicinctus, etc) or at all (e.g., ratios of eye 
length to head width and head width to neck width, and aspects of 
colour pattern). In those cases, specimens at hand were examined to 
fill in the gaps. I have examined preserved material of most taxa, and 
have observed and collected 12 of the species: T. canus, T. feicki, T. 
fuscus, T. greenwayi, T. haetianus, T. maculatus, T. melanurus, T. 
pardalis, T. pilsbryi, T. stejnegeri, T. stullae, and T. wrighti. 

Because this is not a comprehensive revision, there was no 
attempt to survey all collections for holdings of Tropidophis or to 
examine all available material. It is anticipated that such an under- 
taking will be attempted in the future. 



RESULTS AND DISCUSSION 

The conclusion of this taxonomic update is the recognition of 29 
species of Tropidophis (Table 1). This is an increase of about six 
species over the number recognised earlier this year (Hedges & 



Garrido, 2002). The difference involves the elevation of some taxa 
previously considered as subspecies. Below, I discuss the utility of 
different characters used, my reasoning in determining species 
boundaries, and the taxonomic issues involved in each geographic 
area. The phylogeny and biogeography of species in this genus, 
using DNA sequence data, is discussed elsewhere (S. B. Hedges. S. 
C. Duncan, A. K. Pepperney, in preparation). The species group 
status (Table 1 ) is based on that work, but otherwise the focus of this 
current assessment is the definition of species boundaries, not 
phylogenetic relationships. 

Characters 

Variation in 20 characters among the 29 species of Tropidophis is 
shown in Tables 2-4. They are grouped into those involving propor- 
tions (Table 2), scalation (Table 3), and pattern and coloration (Table 
4). In general, sexual dimorphism in Tropidophis is not pronounced 
and therefore data from both sexes can be combined, with the 
exception of body size, which shows slight differences. Characters 
that I have found to be of limited value have been eliminated. These 
include four that are commonly scored in snake systematics: upper 
and lower labials and the pre- and postoculars. All four are variable 
within species and in almost all cases, not diagnostic. Upper labials 
are usually 9-10 and lower labials usually 9-12 in all species. In T. 
melanurus and some related species, labial counts tend to be higher, 
although even in those cases there is often overlap. There is usually 
one preocular and 2-3 postoculars in Tropidophis, although some 
species occasionally have two preoculars and as many as 4 
postoculars; however, variation in ocular scales does not appear to 
be of taxonomic utility. Examples of exceptions, as noted by Schwartz 
and Marsh (1960), are T. pardalis (usually 2 postoculars) and T. 
maculatus (usually 3 postoculars), although such differences are 
rarely diagnostic. Stull ( 1 928) considered the forking of the hemipenis 
(bifurcate versus quadrifurcate) to be a diagnostic character but 
Schwartz and Marsh (1960) could not identify any species or 
specimens with a quadrifurcate condition. Also, such a character 
would not be very useful in this group because of limited material 
and scarcity of specimens with properly everted hemipenes. 

Schwartz scored several other characters in Tropidophis, but I 
have also found them to be of limited value in diagnosing taxa. In the 
case of relative tail length (Schwartz & Marsh, 1960), it is useful in 
distinguishing T. canus from T curtus (see below) but otherwise is 
difficult to score because of tail damage in some specimens, and 
overlapping of ratios. The colour of the tail tip (pale versus dark) was 
useful in distinguishing Cayman Islands Tropidophis from T. 
melanurus (Thomas, 1963), and other trends are noticeable, but 
differences between juveniles and adults, and intraspecific variabil- 
ity, make it a less useful character. 

Now considering the 20 tabulated characters, maximum snout- 
vent length (SVL) is useful because some species differ greatly in 
body size, and most individuals encountered are adults. Two ratios 
(Table 2) that I have found to be of utility are eye length/head width 
(i.e., relative size of the eye) and head width/neck width (i.e., 
distinctiveness of the head). Both ratios are larger in the arboreal 
species T feicki, T semicinctus, and T. wrighti, and in another 
gracile Cuban species (T. fuscus) that is possibly arboreal (Hedges & 
Garrido, 1992). Unfortunately, both show variation within species 
and sample sizes still are small. 

Despite the intraspecific variability in the scale characters (Table 
3), some are useful when considered simultaneously with other 
characters. Ventral and midbody scale row counts are perhaps the 
most useful whereas caudal counts and posterior scale row counts 
are the least useful. Contact of the two parietal scales can be 






SNAKE OF THE GENUS TROPIDOPHIS 



85 



Table 2 Variation in proportions of snakes of the genus Tropidophis. 





Max 


SVL(mm) 


Eye diameter/ 


Head width/ 






Species 


males 


females 


head width 


neck width 


Sample size 1 


References 2 


T. battersbyi 


na 3 


na 


na 


na 


1 


1 


T. bucculentus 


360 


596 


0.19-0.24(2) 


1.50-1.55(2) 


4 


2-5 


T. canus 


363 


338 


na 


na 


20 


2,6 


T. caymanensis 


470 


438 


0.26(1) 


1.59(1) 


13 


2-3. 7 


T. celiac 


na 


344 


0.28(1) 


1.31(1) 


1 


8 


T. curtus 


357 


354 


0.25(1) 


1 .35 ( 1 ) 


93 


2-3, 6 


T. feicki 


411 


448 


0.28-0.32 (4) 


1 .76-2.24 (4) 


29 


2-3.9 


T. fitscus 


287 


304 


.30-.33 (2) 


1.83-1.99(2) 


8 


10-11 


T. galacelidus 


187 


405 


0.28 ( 1 ) 


1.45(1) 


6 


2-3.6, 12 


T. greenwayi 


313 


301 


0.23 ( 1 ) 


1.35(1) 


16 


2-3.6 


T. haetianus 


534 


552 


0.22-0.25 (8) 


1.28-1.52(8) 


158 


2-3,6, 13 


T. hardyi 


303 


334 


0.26-0.31 (2) 


1.30-1.49(2) 


8 


2-3.6, 12 


T. hendersoni 


302 


315 


0.28 ( 1 ) 


1.45(1) 


1 


14 


T. jamaicensis 


338 


306 


0.20-0.21 (3) 


1.47-1.54(3) 


23 


2-3,6 


T. maculatus 


327 


347 


0.23-0.32 (5) 


1.30-1.92(5) 


25 


2-3,6 


T. melanurus 


770 


957 


0.21-0.26(8) 


1.28-1.77(8) 


100 


2-3, 15 


T. morenoi 


na 


295 


0.24-0.27 (2) 


1.39-1.52(2) 


2 


16 


T. nigriventris 


184 


227 


na 


na 


4 


2.6. 12 


T. pardalis 


264 


287 


0.24-0.27 (4) 


1.26-1.63(4) 


161 


2-3.6 


T. parkeri 


422 


512 


0.24 ( 1 ) 


1.95(1) 


21 


2-3, 7 


T. paucisquamis 


101 


283 


0.24-0.28 (3) 


1.53-1.71 (3) 


3 


2-3 


T. pilsbryi 


295 


260 


.24-.25 (2) 


1.59-1.62(2) 


8 


2-3,6, 10 


T. schwartzi 


385 


321 


na 


na 


17 


2-3. 7 


T. semicinctus 


383 


408 


0.30-0.34(2) 


1.70-1.88(2) 


26 


2-3,9 


T. spiritus 


320 


372 


0.24-0.37 (4) 


1.35(1) 


4 


17 


T. stejnegeri 


395 


529 


0.22-0.28(3) 


1.39-1.48(3) 


23 


2-3.6 


T. stullae 


260 


248 


0.23-0.25(3) 


1.78-1.86(3) 


4 


2-3.6 


T. taczcmowskyi 


305 4 


243 


0.27-0.30 (2) 


1.46-1.51 (2) 


3 


3, 10, 18 


T. wrighti 


330 


323 


0.32-0.34 (7) 


1.77-2.24(7) 


17 


2-3,9 



'number of specimens used for most measurements and counts, unless otherwise indicated in parentheses. 

: primary sources of the data reported in this and other tables: I (Laurent, 1949). 2 (Albert Schwartz, unpublished data). 3 (S. B. Hedges, unpublished data), 4 (Thomas. 1966). 5 

(Bailey, 1937), 6 (Schwartz & Marsh. I960). 7 (Thomas, 1963), 8 (Hedges el ui, 1999). 9 (Schwartz. 1957). 10 (Hedges & Garrido. 1992). 1 I (Ansel Fong. unpublished data), 

12 (Schwartz & Garrido, 1975). 13 (Schwartz. 1975). 14 (Hedges & Garrido. 2002). 15 (Schwartz & Thomas. I960). 16 (Hedges et al., 2001), 1 7 ( Hedges & Garrido. 1999), 18 

(Stull, 1928). 

'data not available 

4 sex not determined 



diagnostic in some comparisons (Hedges & Garrido, 2002). but 
problems arise in how different people score the character (e.g., 
when an interparietal is present and scales barely touch). As already 
noted, the keeling of the dorsal scales is often variable within 
species. Many species have weakly keeled scales that are noticeable 
only above the vent region and are difficult to score consistently, and 
depend sometimes on condition of preservation. However, some 
species consistently have smooth scales and others (e.g., T. 
melanurus) have distinctly keeled scales. 

Colour and pattern variation (Table 4) has been important in 
Tropidophis taxonomy, in part because the snakes are frequently 
spotted and this provides yet additional characters to count. In fact. 
Schwartz and Marsh ( 1960) considered coloration and pattern to be 
the most reliable characters, in combination with scalation, for 
'separating and combining' taxa. Except for T. feicki, which has 
crossbands, most species have 2-12 rows of body spots. I have used 
the Schwartz and Marsh (1960) methods of scoring body spots and 
spot rows. Spot rows include those on the dorsum and venter, all 
around the body (both sides) whereas body spots are counted along 
one row of spots (usually just to one side of middorsal region) from 
behind the head to just above the vent. Typically, the largest and 
most distinctive spots are those near the middorsal region. This 
reaches an extreme in species of the melanurus group where some 
individuals have only those two spot rows present, resulting in 
widely varying row counts (e.g., 2-10). Occipital spots sometimes 
fused to form a white neckband, are diagnostic of several species 
(e.g., T. celiae, T. galacelidus, T. pilsbryi, T. stejnegeri) and are 



common in others (e.g., T. pardalis). 

The dorsal ground colour of most species is a shade of brown or 
grey, and often variable within species. I once collected two speci- 
mens of T. pilsbryi in the same rock pile in Cuba, and was initially 
misled into thinking they were different species because one was 
brown and the other grey. On the other hand, T. stullae is consist- 
ently pale tan and differs from the other two Jamaican species, which 
are darker. Also, two boldly spotted species that occur sympatrically 
in western Cuba can be distinguished by, among other things, their 
dorsal ground colour: greyish pink in T. feicki and yellow to orange 
in T. semicinctus. Although most species are spotted, those in the 
melanurus group often have narrow lateral stripes as well as a 
middorsal stripe. The absence of middorsal spot contact occurs in 
two related species, T. maculatus and T. semicinctus, and the two 
Bahaman species T. canus and T. curtus are united by the presence 
of an anteriolateral (face and neck) stripe. Ventral pattern is diagnos- 
tic for T. nigriventris (almost completely dark) and in several species 
that lack a ventral pattern, but otherwise most have different degrees 
of spotting and flecking. 

Species boundaries 

Most taxonomists discern the presence of sympatric species by 
covanation of multiple characters from individuals of a single 
locality, indicating lack of gene flow between the species. For 
example, in a series of dark and pale snakes found together, two 
species would be indicated if all of the dark snakes also had small 



86 














S.B. HEDGES 


Table 3 Variation 


in scalation of snakes of the genus 


Tropidophis. (Numbers 


and character states in 


brackets represent rare or infrequent occurrences; Y = 


yes, N = no; other notation as in Table 2. 






















Dorsal scale rows 




Parietal 


Keeled 


Species 


Ventral s 


Caudals 


Anterior 


Midbody 


Posterior 


contact 


dorsals 


T. battersbyi 


200 


41 


21 


23 


17 


N 


N 


T. bucculentus 


183-186 


28-32 


24-25 


25-27 


17-19 


N 


Y 


T. canus 


170-183 


29-35 


21 [20.22,23] 


23[22] 


16-21 


N/Y 


Y[N] 


T. caymanensis 


1 83-200 


33-38 


23-27 


23[25] 


17[19] 


N 


N/Y 


T. celiae 


203 


30 


25 


27 


19 


Y 


N 


T. curtits 


146-173 


22-37 


19-27 


23-25 


17-22 


N[Y] 


Y[N] 


T. feicki 


217-235 


34^11 


23-25[19,21] 


23-25 


17-19 


N/Y 


N 


T. fuscus 


160-185 


30-36 


21-24 


23 


15-19 


N 


Y 


T. galacelidus 


177-186 


29-35 


25-27 


25-27 


19-20 


N 


Y 


T. greenwayi 


155-165 


26-30 


23-25 


25-27 


17-19 


Y 


N 


T. haetianus 


170-194 


27-39 


23-27 


25-27[23,29] 


17-19[21] 


Y[N] 


N 


T. hardyi 


153-172 


31-48 


20-24 


23-25 


18-20 


N/Y 


N/Y 


T. hendersoni 


190 


33 


23 


25 


19 


N 


Y 


T. jamaicensis 


167-181 


28-36 


23-27 


25-29 


15-23 


N/Y 


N 


T. maculatus 


189-208 


28-40 


22-25 


25[23] 


17-21 


N/Y 


N/Y 


T. melanurus 


188-217 


31^44 


24-27 [19] 


27-29 
[24,25,26,30] 


17-21 

[16,22,23,24] 


N 


Y 


T. morenoi 


198-199 


42-44 


23 


23 


17 


N 


N 


T. nigriventris 


144-150 


25-26 


23-25 


23-25 


18-22 


N 


N 


T. pardalis 


140-157 


23-34 


21,23 
[19,22,24,25] 


23,25 
[21,22,24] 


1 7-2 1 [ 1 6] 


N/Y 


N[Y] 


T. parked 


199-212 


33^41 


25[23,24] 


27[25,26] 


17[18.19] 


N 


Y 


T. paucisquamis 


170-178 


37-40 


21 


21 


17 


Y 


N 


T. pilsbryi 


160-169 


26-31 


22-25 


23-25 


17-21 


N 


N/Y 


T. schwartzi 


191-205 


31-39 


25 


25[26] 


17[15] 


N 


Y 


T. semicinctus 


201-223 


33-41 


21.23[22,24,25] 


25 [2 1-24] 


17-20 


N/Y 


N 


T. spiritus 


183-200 


35-39 


21-23 


23 


17 


N 


N 


T. stejnegeri 


181-190 


30-38 


25-27[23] 


25,27[26] 


17-19 


N/Y 


Y 


T. stullae 


166-170 


31-34 


25 


25 


16-19 


N 


N 


T. taczanowskyi 


149-160 


25-27 


23-25 


23 


19-21 


Y 


Y 


T. wrighti 


192-215 


36-45 


21-23 


21-23 


17[16.18,19] 


N 


N 



heads and fewer spots than the pale snakes (thus, body colour would 
be covarying with head size and spot number). In the case of 
allopatric populations, it is typically assumed that character differ- 
ences similar to or greater than observed between sympatric species 
indicate that the two forms are different species. Thus, the 'yard- 
stick' used for assessing allopatric populations is character divergence 
between closely related, sympatric species. This is the principle that 
I use here in assessing species status within Tropidophis. It is a 
practical species concept but is based on the observation that species 
are reproductively isolated from each other, as noted by Darwin 
( 1 859) and later articulated by Mayr ( 1 942) as the biological species 
concept. 

The reason that a particular degree of differentiation is necessary, 
rather than a minimal diagnostic difference, concerns the 'reality' of 
species in evolution. Almost all species are fragmented (structured) 
to some degree, and many populations can be diagnosed by one or a 
few nucleotide differences or minor morphological differences. 
However, through time, such populations frequently combine and 
separate again as part of the reticulate nature of gene flow and 
evolution within species. It is only those populations that have 
differentiated sufficiently, genetically and/or morphologically, and 
presumably reflecting a length of time, that evolve reproductive 
isolation from other populations. Thus, to assign species status to 
diagnosable, but ephemeral, populations during one slice of time is 
arbitrary from an evolutionary standpoint. Although Frost and Hillis 
(1990) recommended abandoning the use of quantitative criteria 
(molecular and morphological) for discerning species status of 
allopatric populations, they did not propose anything to replace that 
procedure and thus few have heeded their recommendation. 



Sympatric species of Tropidophis occur only in Cuba. In western 
Cuba, the following six species have been found in the general 
region of Canasf, Habana Province: T. celiae, T. feicki, T. maculatus. 
T. melanurus, T. pardalis, and T. semicinctis. In central Cuba, the 
following six species have been found in the vicinity of the Trinidad 
mountains: T. galacelidus, T hardyi, T melanurus, T. pardalis, T. 
semicinctis, and T spiritus. In eastern Cuba, the following four 
species are known from the region of Baracoa, Guantanamo Prov- 
ince: T. fuscus, T. melanurus, T pilsbryi, and T. wrighti. To identify 
the level of character divergence associated with species differentia- 
tion in Tropidophis, I now focus on four clusters of sympatric 
species, each of which are members of the same species group: ( 1 ) 
feicki/maculatus/semicinctis,(2) celiae/melanurus, (3) pardalis/ 
galacelidus/hardyi, and (4) fuscus/wrighti/pilsbryi. 

In cluster ( 1 ), T maculatus and T. semicinctis are closest relatives 
according to DNA sequence evidence (S. B. Hedges, S. C. Duncan, 
A. K. Peppemey, in preparation) and are distinguished primarily by 
colour pattern: the number of body spots (no overlap) and number of 
spot rows (no overlap). All scale counts in those two species overlap, 
although T. semicinctis tends to have a higher number of ventrals. In 
the case of T. feicki and T. maculatus, there are non-overlapping 
differences in ventral counts, body spots, and spot rows. Consider- 
ing T. feicki and T semicinctis, the ground colour and spot rows are 
non-overlapping, and the ventral counts are different but overlap 
slightly. 

In cluster (2), T. celiae and T. melanurus, which are close relatives 
according to DNA sequence evidence, completely overlap in all 
scale counts, although parietal contact might be considered diagnos- 
tic if there were more than one specimen of T. celiae. Otherwise. 



SNAKE OF THE GENUS TROPIDOPH1S 



87 



i^ 2 



a o 

^ a. 



- 













<1> 








B 


- 


< 


— 





o 


— 


— 




o 








-a 


— 


•-J 


'J 




T3 








u 


<u 


-£ 


^: 


V 


U 


<u 


X 


X s: 






CJ 


o 














u 


-J 


c 





B 


; 


o o 


c^ 


c 


c_ 


Q. 


C 


CU 


C 


a 


p. a. 



2 a. 



C oo o "O 

u ecu 

o-ooo.ol.oi> 



v £ '£ '£ £ w 



O o u S ° ° 
a. a. a. 2 o. o. 



§>>->->->->->->->->->->->-Z>->->->->- g >- >- Z >- >- >- 



ZZ>->^ZZ^ZZZZZZ>-ZZZ>-ZZ>->-Z>-> 



zz>-zz>zzzzzzzzzzzzzzzzzzzzzzz 



1) 1) 
a. a. 
■C 'C 



'. 


". 





o 


3 


". 


c_ 


'_ 


'— 


Q. 


C_ 


a. 


C/3 


v: 


r J> 


x 


y: 


^ 







j^ 












a. 






o 




























E "O 






X) 












S C 






-a 


















c 












u 




a 












B J 


-3 




x x 












x o 


















•o 


a 


o o 


o 


C 


3 


3 


3 


Q. B 


C3 


c^ 


o. c 


a 


Q. 


a. 


a. 


Q. 


C/3 


x 


•A s 


A 


X 




(/J 


X 


y C 



cA ca ca ua 



<Ji s: si t/: c/5 



oooooooooooo 

s xtcacacococAcacacowcfl 



is E 

o n 

c = Si 

£ u o 

O ^"O 

g X> V! C3 

O JZ 1) u 

oo oo 2 



P C 



4) <u (U u dj >•»>-* .^ >^_i^ 

~ -t; — ~= ~z '-> '-> = ij = 



rt 3 C3 73 77J 



cu cu 



o c 

e B 



3 r 



S3 o 



-3 '!-» — ^ 



o aj 



»« t^ W W J ' — _ „ _ . U. .— 

0- Q. a. O. O. 00 00 T3 OJ0T3X c 



o «: 
2 S 



S 2 

S c 

— 3 



&| 

u 5 
n Si 

Q. CD 



5 ^ 

> c 

O a 

So 






-X J-> ■ — ; tU 



>^ Oij >" Q.-D 



P -C 



zzz>-zzC>. Z zz>-zz 



z^z 



>-zzz>-zzz 



o ^ o o o o ^ 

.' T 22, i °f T i i o 

vO VD vO (N vO - — sO — — 0O\D — vOOOOI\000\0(N\000<N(NvOOO — OO^t 



|-~ r— O — VD 
I I I CM I I 

in m vi — <n (N 



O - — n - 

a\ \c ~— vp oo — -- r^ooinr-^o oo on On ^o ^- oo 

O 4- u-i m m -4- in 






*ON- 

u-> u^ vO 
I I I 

« oo m oo o oo _ 
c t l-Nt^on- 



o , orJOO\-NtcN\t | ONtC\^ONrJ\Ooo'to\(Nt^Nt 
iiiiiTliiiiilliiliilii 

" ^rr-\OfNooooi/ir--oor-~ino>n^DONOoo\Or<-, 



fi> <j a fc .a a ~>: 



? c ^ 



^ ^ 'r- "^ ^e* ^ -d 

-^ r- :-. i^ ^ ^ ^ 






Q ^ O.^ 



'S 'C .£ S 8 * 






•o ^^e, oo oojj -c j: -S, e 5 S e cn,ci. 



88 



S.B. HEDGES 



about the only characters that distinguish these two species are body 
size and aspects of coloration (e.g., neckband in T. celiae and higher 
number of body spots). In case the reader is wondering, the presence 
of enlarged ova in the small holotype of T. celiae, and details of the 
pattern, indicate it is not a juvenile T. melanurus (Hedges et ah, 
1999). 

In cluster (3), there are no molecular data available for T. 
galacelidus and T. hardyi to confirm their species group association 
with T pardalis. However the association is supported by the fact 
that there are no diagnostic (non-overlapping) scale or pattern 
characters that distinguish T. hardyi and T. pardalis. This problem 
was noted in the original description (Schwartz & Garrido, 1975). 
However, T. hardyi has a higher number of ventrals, even though 
overlapping with T. pardalis, and it is a larger species with a 
distinctly smaller head. The latter character caused Schwartz and 
Garrido to associate (as a subspecies) T. hardyi with the small- 
headed T. nigriventris. The third sympatric species of this trio, T. 
galacelidus, can be distinguished from the other two species by its 
higher number of ventrals, dorsal spots, and spot rows (all non- 
overlapping). 

In the case of cluster (4), DNA sequence evidence place all three 
together as close relatives. Tropidophis fuscus and T. pilsbryi have 
no completely diagnostic scale differences, although the combina- 
tion of ventral scale counts and midbody scale rows will distinguish 
the species. Also, T. fuscus has a more gracile body shape. The third 
species, T. wrighti, is diagnosed from the other species by its higher 
ventral counts, and fewer dorsal spots and spot rows (all non- 
overlapping). 

To summarize, of the ten combinations of closely related, sympatric 
species, nearly all were distinguished by at least two non-overlap- 
ping differences in colour pattern, or (less frequently) body 
proportions. In addition, there was usually one other difference 
(either non-overlapping or overlapping) in scalation. More distantly 
related species of Tropidophis often have two (or more) non-over- 
lapping differences in scalation, in addition to any other differences. 
This suggests a temporal sequence in character differentiation, with 
colour pattern and body proportion differences accruing first, fol- 
lowed by scalation differences. Ideally, one would like to use 
molecular data as well for assessing differentiation, although tissue 
samples still are not yet available for many taxa. Using this morpho- 
logical criterion for assessing species status in Tropidophis, I will 
now review the current status of the taxa in this genus. 

Hispaniola 

Only one species (T haetianus), with three subspecies, occurs on 
Hispaniola: T h. haetianus (most of island), T h. hemerus (distal 
portion of the Tiburon Peninsula in Haiti) and T. h. tiburonensis 
(extreme eastern portion of the Dominican Republic). Although 
Schwartz and Marsh (1960) and Schwartz (1975) have considered 
the Jamaican taxa to be subspecies of T haetianus, genetic evidence 
has shown that they are more closely related to the Cuban species 
(Hass, Maxson & Hedges, 2001) and thus are removed from T 
haetianus (see below). Also, the Cuban specimens of T haetianus 
discussed by Schwartz and Marsh (1975) and Schwartz and Garrido 
(1975) have been removed from that species and assigned to a new 
species, T hendersoni (Hedges & Garrido, 2002). Because the 
subspecies of Hispaniolan T haetianus are parapatric and apparently 
intergrade (Schwartz, 1975), and because their character differentia- 
tion is less than that of sympatric species, I suggest retaining their 
current taxonomic status as subspecies. It is possible that genetic 
studies in the future may further clarify their status. Thus, T. 
haetianus is confined to Hispaniola and contains three subspecies. 



Navassa Island 

Four specimens of T bucculentus are known from this small island 
between Hispaniola and Jamaica, but apparently no snakes have 
been seen in over 1 00 years and thus the species is considered extinct 
(Powell, 1999). Since it was described by Cope (1868), there has 
been considerable confusion as to its species status and relationship 
with other species. Most who have examined the type series, includ- 
ing me, have noted a resemblance to T melanurus (Thomas, 1966), 
although Stull ( 1 928) instead considered it a subspecies of T pardalis. 
There is no overlap in ventral counts between T bucculentus and T 
melanurus, and almost no overlap in caudal counts. Although there 
appear to be pattern differences between the two species, the single 
specimen in the Academy of Natural Sciences (Philadelphia) differs 
from the other three specimens (National Museum of Natural His- 
tory, Smithsonian) in terms of ventral pigmentation (Bailey, 1937). 
Based on the diagnostic scalation differences alone, I would con- 
sider T. bucculentus as a valid species. The unusual geographic 
location of a species with apparent Cuban affinities, on Navassa 
Island, is remarkable. With the exception of the anole (Anolis 
longiceps), other species on Navassa have affinities with nearby 
Hispaniola (Powell, 1999; Thomas, 1966), which is logical based on 
the westerly direction of ocean currents. However, the eastern tip of 
Cuba is further east than Navassa. and ocean currents flow southerly 
through the Windward Passage separating Cuba and Haiti. Dispersal 
on those currents is thus possible and is the most likely explanation 
for the origin of T bucculentus (and A. longiceps) on Navassa and 
possibly the gecko Sphaerodactylus notatus on the Morant Cays 
southeast of Jamaica. The locally changing direction of water 
currents during a hurricane may also have aided in the dispersal of 
these taxa. 

Cuba 

With 15 described species. Cuba is the hot spot of species diversity 
in the genus. Recently, two subspecies described by Schwartz and 
Garrido (1975) were elevated to species status and a new species 
was described from eastern Cuba (Hedges & Garrido, 2002). 
Character differences among many of the Cuban species have been 
discussed above (see 'Species Boundaries'), and I consider all 15 
species to be valid. Also, I am aware of material that likely repre- 
sents additional, undescribed species. Undoubtedly, more species 
will be discovered. 

Two remaining taxa are considered subspecies of T. melanurus: T. 
m. dysodes and T m. eriksoni (Schwartz & Thomas, 1960). The 
former is known from three female specimens from near La Coloma, 
Pinar del Rio Province, and the latter is restricted to Isla de Juventud. 
These taxa differ from T. m. melanurus primarily in size of the dorsal 
spots and in having bolder, darker colouration, with T. m. dysodes 
having the darkest pigmentation of the three subspecies. The ventral 
counts of T. m. eriksoni are low for the species, but there is 
considerable overlap with the other two taxa. Considering that there 
are no diagnostic differences in body proportion or scalation, and the 
colouration differences, although real, are not as trenchant as those 
distinguishing sympatric. closely related species (e.g., T maculatus 
and T. semicinctis), I am inclined to leave their status as subspecies 
unchanged until additional data warrant a reconsideration. 

Jamaica 

The three Jamaican taxa, originally described as full species, are 
closer to Cuban taxa than to T. haetianus based on immunological 
data (Hass et al, 2001) and DNA sequence data (S. B. Hedges, S. C. 
Duncan, A. K. Pepperney. in preparation). However, they form a 



SNAKE OF THE GENUS TROP1DOPHIS 



89 



single genetic and morphological group (jamaicensis group), and 
are distinguished morphologically from the Cuban species at the 
species level, although they are closest to species of the pardalis 
group. The question then remains as to whether they should be 
treated as a single species (T jamaicensis) or three separate species: 
T. jamaicensis, T. stejnegeri, and T. stullae. However, using the 
morphological criterion for species status, I recommend the latter. 
Each of these three taxa can be diagnosed based on scalation, body 
proportions, and colour pattern, and they are as different from each 
other as sympatric species in Cuba. In body size, T. stejnegeri (529 
mm S VL) is considerably larger than T. stullae (260 mm S VL), with 
T. jamaicensis (338 mm SVL) being intermediate in size. Ventral 
counts of T. stejnegeri do not overlap with those of T. stullae, and 
counts of T. jamaicensis are nearly completely non-overlapping 
with the other two taxa. Tropidophis stejnegeri has keeled scales and 
occipital spots whereas the other two taxa are smooth scaled and 
lack occipital spots. Additionally, dorsal ground colours differ, 
being yellowish-grey (T. stejnegeri), chocolate brown (T. 
jamaicensis) and pale tan (T stullae). A middorsal stripe is present 
in T. stejnegeri and T. stullae but absent in T. jamaicensis. The head 
of T. stejnegeri is pointed but that of T. stullae is distinctly squared- 
shaped. 

The Bahamas Bank 

Six taxa are currently recognized from the Bahamas Bank: 
Tropidophis canus androsi Stull (Andros Island), T. c. barbouri 
Bailey (central Bahamas, from Eleuthera to Ragged Island), T. c. 
canus Cope (Great Inagua), T. c. curtus Garman (New Providence, 
Bimini Islands, and Cay Sal Bank), T. g. greenwayi Barbour and 
Shreve (Ambergris Cay), and T. g. lanthanus Schwartz (Caicos 
Islands). Schwartz and Marsh (1960) considered all except the last 
two to be subspecies of a single species (T. canus) and that arrange- 
ment has since been followed. However, it is worth reviewing 
morphological variation in T. canus in the context of our current 
understanding of species definitions in the genus. Recent evidence 
from DNA sequences has shown that T greenwayi is most closely 
related to T. haetianus (Hispaniola) and unrelated to the complex 
currently considered under T. canus. 

Among the four subspecies of T. canus, T. c. canus stands out both 
morphologically and geographically. It is isolated in the south, being 
separated from the northern taxa by islands apparently lacking 
Tropidophis: Crooked, Acklins, Mayaguana, and Little Inagua. It 
has a higher number of ventrals (170-183). One specimen (1%) of 
the northern group has 173 ventrals; all others have fewer than 168 
ventrals. Anterior and midbody scale rows in T c. canus typically 
are 21-23 whereas they are typically 23-25 in the northern taxa, 
although there is some overlap. The tails of T. c. canus are distinctly 
shorter, averaging 11% (9.4-12.1), compared with 13% (11.0-15.2) 
in the northern taxa. Rows of body spots number 6-8 in T. c. canus 
whereas they are typically 10 or more in the northern taxa; overlap 
consists of nine specimens (10%) of northern taxa with eight rows 
and two (2%) with nine rows, and one (5%) T. c. canus with nine 
rows. This degree of difference is the same or greater than that seen 
between sympatric species of Tropidophis in Cuba, and therefore the 
northern taxa should be removed from T canus. 

The status of the three northern Bahaman taxa is problematic at 
this time. Clearly there is geographic variation among these forms. 
For example, androsi tends to have a higher number of ventral scales 
than the other two taxa, although there is considerable overlap with 
barbouri and some with curtus. Within one taxon (curtus), snakes 
from Bimini are distinctly larger than those from New Providence. 
Both Bailey (1937) and Schwartz and Marsh (1960) noted very little 



difference, overall, between barbouri and androsi. When consider- 
ing the 'species boundary' characters noted above, there is insufficient 
justification at present to recognize these taxa as distinct species. 
Additional specimens and genetic analyses will be necessary to 
better resolve geographic variation in northern Bahaman Tropidophis. 
Until then, I suggest here that androsi and barbouri be recognized as 
subspecies of T curtus: T. curtus androsi (new combination) and T. 
curtus barbouri (new combination). 

Tropidophis greenwayi lanthanus is a subspecies found in the 
Caicos Islands and is distinguished by coloration difference from the 
nominate subspecies on nearby Ambergris Cay (Schwartz, 1963). 
However, the difference concerns 'interspace stippling' and not 
actual numbers of spots or spot rows. There are no diagnostic scale 
count differences, and the presence of two postoculars in the two 
known specimens of T g. greenwayi is not remarkable because half 
of the specimens of T. g. lanthanus also have two postoculars, at 
least on one side of the head. More material of T. g. greenwayi is 
needed, in addition to genetic analyses, before the species status of 
T g. lanthanus can be accurately assessed. I suggest that the latter 
taxon continue to be recognized as a subspecies. 

Thus, Tropidophis of the Bahamas Bank are placed here in three 
species: T. greenwayi (Turks and Caicos), T. canus (Great Inagua), 
and T. curtus (northern and central Bahamas). The question as to 
whether some Bahaman species also occur in Cuba has been raised 
in the past, primarily because of two old specimens (Schwartz & 
Marsh, 1960). The first is the type of T. curtus, purportedly from 
'Cuba' (Garman, 1887). However, morphologically it agrees with 
snakes from New Providence, Bahamas, and the specimen number 
(MCZ 61 14) is close to other numbers in that collection from New 
Providence. Also, the origin of the specimen was investigated and 
found to be 'without definite history' (Stull, 1928). Thus, I agree 
with Stull in considering this specimen to be from New Providence. 
The other specimen is AMNH 2946 from 'Nuevitas, Cuba' (no other 
information). As noted by Schwartz and Marsh (1960) it agrees in 
morphology with snakes here considered as T curtus. Although they 
considered the provenance of the specimen to be correct, partly 
because of the confusion surrounding the holotype, I raise the 
question here that it also may be an error. The specimen number is 
close to several T. curtus from Andros Island (AMNH 2925-2927) 
apparently cataloged at about the same time and its scale counts fall 
within the range of counts of snakes from that island. Thus I consider 
the range of T curtus to be restricted to the Bahamas. 

The Cayman Islands 

Currently there are three subspecies of T. caymanensis recognized 
from the Cayman Islands (Thomas, 1963) and they differ in scale 
row counts, ventral counts, and colour pattern. Each is endemic to a 
single island, and there is no evidence of intergradation. At the time 
they were last reviewed (Thomas, 1963), a more conservative 
definition of species boundaries in the genus prevailed. Although no 
new material has been examined here, the level of differences seen 
among these taxa would suggest that they are distinct species. 
Tropidophis caymanensis (Grand Cayman) is distinguished from T 
parkeri (Little Cayman) by its lower anterior and midbody scale 
rows (23-25 versus 25-27), lower number of ventrals (183-200 
versus 199-212), and a larger, darker cephalic pattern. Tropidophis 
caymanensis is distinguished from T schwartzi (Cayman Brae) by 
its larger body size (maximum SVL = 470 mm versus 385 mm), 
lower anterior scale rows (23 versus 25), lower, albeit overlapping, 
number of ventrals (183-200, x = 192, versus 191-205, x = 198), 
fewer tail spots (4-8, mode = 6 versus 5-9, mode = 8) and a larger, 
darker, cephalic pattern. Tropidophis parkeri is distinguished from 



90 



S.B. HEDGES 



T. schwartzi by its higher midbody scale rows (27 versus 25), higher 
number of ventrals (199-21 2, x = 203 versus 191-205, x = 198), and 
a larger, darker cephalic spot (Thomas, 1963). 

South America 

Although Stull (1928) and Schwartz and Marsh ( 1960) attempted to 
relate one or more of the South American taxa to West Indian species 
groups, I do not envision a close relationship. For example, the 
keeling of the dorsal scales in T. taczanowskyi is greater than I have 
seen in any West Indian taxon. In the case of T. paucisquamus, the 
low number (21) of midbody scale rows and a distinctive pattern of 
middorsal stripe and blotches is not like any West Indian species, as 
noted by Schwartz and Marsh (1960). The only known specimen of 
T. battersbyi has been described only as having six rows of spots, 
including two rows on the venter (Laurent, 1949; Perez-Santos & 
Moreno, 1991). The fact that the venters of T. paucisquamus and T. 
taczanowskyi have both been described as consisting of black and 
yellow spots and bands (Stull, 1928) is noteworthy; such a pattern 
and colouration is not known in West Indian taxa. This might also 
suggest a relationship at least between these two species. Molecular 
phylogenetic evidence (S. B. Hedges, S. C. Duncan, A. K. Pepperney, 
in preparation) places T. paucisquamus outside of the West Indian 
clade, reinforcing the morphological distinction. Examination of 
additional specimens, and genetic data from T. battersbyi and T. 
taczanowskyi, are needed to clarify the relationships of these South 
American species. Until then, available evidence supports the place- 
ment of the South American species in a separate species group 
(taczanowskyi group). 



Acknowledgements. I thank R. Henderson for providing access to the 
raw scale count data and notes of A. Schwartz; L. Diaz, A. R. Estrada, A. 
Fong, O. H. Garrido, and L. Moreno, for data on specimens in their posses- 
sion; R. Thomas for assistance in the field; the staffs of the National Museum 
of Natural History (Smithsonian), Museum of Comparative Zoology 
(Harvard), and Natural History Museum (London), for loan of material or 
access to the collections. This work was supported by grants from the U.S. 
National Science Foundation. 



REFERENCES 

Bailey, J.R. 1937. A review of some recent Tropidophis material. Proceedings of the 

New England Zoological Club 16: 41-52. 
Boulenger, G.A. 1893. Catalogue of snakes in the British Museum (Natural History). 

Vol. I. Longmans and Company, London. 



Cope, E.D. 1868. An examination of the Reptilia and Batrachia obtained by the Orton 

Expedition to Ecuador and the upper Amazon, with notes on other species. Proceed- 
ings of the Academy of Natural Sciences, Philadelphia 20: 96-140. 
Darwin, C. 1859. The Origin of Species. John Murray, London. 
Frost, D.R. & Hillis, D.M. 1990. Species in concept and practice. Herpetologica 46: 

87-104. 
Garman, S. 1887. On West Indian reptiles in the Museum of Comparative Zoology. 

Cambridge, Massachusetts. Proceedings of the American Philosophical Society 24: 

278-286. 
Hass, C.A., Maxson, L.R. & Hedges, S.B. 2001. Relationships and divergence times 

of West Indian amphibians and reptiles: insights from albumin immunology. In: 

Woods CA and Sergile FE, eds. Biogeography of the West Indies: patterns and 

perspectives. Boca Raton. Florida: CRC Press. 157-174. 
Hedges, S.B., Estrada, A.R. & Diaz, L.M. 1999. A new snake (Tropidophis) from 

western Cuba. Copeia 1999: 376-38 1 . 
& Garrido, O.H. 1992. A new species of Tropidophis from Cuba (Serpentes. 

Tropidophiidae). Copeia 1992: 820-825. 
& . 1999. A new snake of the genus Tropidophis (Tropidophiidae) from 

central Cuba. Journal of Herpetology 33: 436-441. 
& . 2002. A new snake of the genus Tropidophis (Tropidophiidae) from 

Eastern Cuba. Journal of Herpetology 36: 157-161. 
, & Diaz, L.M. 2001. A new banded snake of the genus Tropidophis 

(Tropidophiidae) from North-Central Cuba. Journal of Herpetology 35: 615-617. 
, Hass, C.A. & Maugel TK. 1989. Physiological colour change in snakes. Journal 

of Herpetology 23: 450-455 
Laurent, R. 1949. Note sur quelques reptiles appartenant a la collection de 'Institut 

Royal des Sciences Naturelles de Belgique. III. Formes americaines. Bulletin Institut 

Royal des Sciences Naturelles Belgique. Bruxelles 25: 1-20. 
Mayr, E. 1942. Systematica and the origin of species. Columbia University Press. New 

York. 
Perez-Santos, C. & Moreno AG. 1991. Serpienles de Ecuador. Museo Regionale di 

Scienze Naturali. Torino, Italy. 
Powell, R. 1999. Herpetology of Navassa Island. West Indies. Caribbean Journal of 

Science 35: 1-13. 
Schwartz A. 1957. A new species of boa (genus Tropidophis) from Western Cuba. 

American Museum Novitates (1839): 1-8. 
. 1963. A new subspecies of Tropidophis greenwayi from the Caicos Bank. 

Breviora. Museum of Comparative Zoology (194): 1—6. 
. 1975. Variation in the Antillean boid snake Tropidophis haetianus Cope. Journal 

of Herpetology 9: 303-31 1. 
& Garrido, O.H. 1975. A reconsideration of some Cuban Tropidophis (Serpentes. 

Boidae). Proceedings of the Biological Society of Washington 88: 77-90. 
& Henderson, R.W. 1991. Amphibians and reptiles of the West Indies. University 

of Florida Press. Gainesville. 
& Marsh, R.J. 1960. A review of the pardalis-maculatus complex of the boid 

genus Tropidophis. Bulletin of the Museum of Comparative Zoology 123: 49-89. 
& Thomas, R. 1960. Four new snakes (Tropidophis. Dromicus. Alsophis) from 

the Isla de Pinos and Cuba. Herpetologica 16: 73-90. 
Stull, O.G. 1928. A revision of the genus Tropidophis. Occasional Papers of the 

Museum of Zoology, University of Michigan: 1—49. 
Thomas, R. 1963. Cayman Islands Tropidophis (Reptilia. Serpentes). Breviora, Mu- 
seum of Comparative Zoology: 1-8. 
. 1966. A reassessment of the herpetofauna of Navassa Island. Journal of the Ohio 

Herpetological Society 5: 73-89. 
Underwood, G. 1967. A Contribution to the Classification of Snakes. British Museum 

(Natural History) Publication No. 653. Trustees of the British Museum (Natural 

History), London. 179pp. 



XX (ZS700U_ 



Bull. nut. Hist. Mas. Loud. (Zool.) 68(2): 91-99 



Issued 28 November 2002 



Atractaspis (Serpentes, Atractaspididae) the 
burrowing asp; a multidisciplinary minireview 



E. KOCHVA 

Department of Zoology, Tel Aviv University, Ramat Aviv, 69978 Tel Aviv, Israel 

CONTENTS 



Dedication 91 

History 91 

Taxonomy 91 

Distribution 92 

Behaviour 92 

Venom apparatus 93 

Venom 93 

Evolution 97 

Acknowledgements 98 

References 98 



SYNOPSIS. The family Atractaspididae is a highly modified derivative of a lineage that apparently arose early in the history of 
'colubroid' snakes, and its taxonomy and relationship with other ophidian groups is still uncertain. Snakes of the genus 
Atractaspis have a characteristic venom apparatus, including the structure and function of the striking unit and of the venom 
glands. The composition of their venom is also unique in containing several low-molecular weight components, thesarafotoxins, 
which affect the cardiovascular system and are similar to the mammalian endothelins. 



Dedication 

This paper is dedicated to Dr. Garth Underwood on the occasion of 
the 35 ,h anniversary of his classic 'Contribution to the Classification 
of Snakes' ( 1967), about which one may say: 

This is a small book by a great man! 

And also -(T ,il ,T"Q) /7J77Z7/7 77A 7 OJJ7H pVUHOJ ItiJT] 
A little that contains a lot (Theodor & Albeck, 1996) 



HISTORY 

There are not many snake species that posed problems from the very 
beginning of their discovery; one of the most prominent ones is 
certainly that which we now call the genus Atractaspis of the family 
Atractaspididae (Fig. 1 ). 

The first two specimens of Atractaspis were described by Reinhardt 
in 1843 as Elaps irregularis, a species that he considered to be 
extremely abnormal because of the presence of only a few, very 
small teeth. On the basis of squamation, Underwood inferred that at 
least one of the specimens was A. dahomeyensis. The genus 
Atractaspis was established by Smith in 1 848 for the South African 
species bibroni and since then it was variously considered as a 
separate family, as a subfamily of the family Viperidae, and finally 
as a genus within the Viperinae. 

Already Haas (1931), on the basis of the pattern of the head 
musculature, was unhappy with the inclusion of Atractaspis in the 
Viperidae, but it was not until 1961 that Monique Bourgeois, 
studying at the Universite Officielle du Congo a Lubumbashi, came 
out with a challenging question: Atractaspis - a misfit among the 



Viperidae? This short note was followed by a detailed study suggest- 
ing the establishment of a separate subfamily for a group of 
opisthoglyphous colubrids together with Atractaspis (Bourgeois, 
1968). 

Underwood (1967) lists a long series of skeletal and other ana- 
tomical characters in which Atractaspis differs from the Viperidae 
and states: '■Atractaspis differs so widely from the other vipers that 
I have no doubts about reviving a separate family group taxon to 
receive it' (p. 103). This he did after a detailed analysis that resulted 
in the resurrection of the subfamily Atractaspidinae (Gunther, 1 858); 
and, finally, the establishment of a separate family. Atractaspididae, 
for the approximately 15 species of Atractaspis together with some 
African colubrid genera (Underwood & Kochva, 1993). 



TAXONOMY 

Recently several suggestions concerning the taxonomy and relation- 
ship of the Atractaspis species have been raised, mainly dealing with 
the question of which additional genera should be included in the 
family Atractaspididae and with which, if any, larger clades they 
should be grouped (Gravlund, 200 1 ). Underwood himself (personal 
communication) is now reconsidering the composition of the family 
Atractaspididae in order to decide which genera, in addition to 
Atractaspis, should be included in it. However, no one is now 
questioning the separate status of the genus Atractaspis, and its 
apparent distinction from the other venomous snake families is 
widely, though not unanimously, agreed upon. Atractaspis thus 
certainly deserves the rank of a family of its own; this may also 
include some rear-fanged snakes that are apparently harmless as far 
as humans are concerned. 



> The Natural History Museum. 2002 



92 



E. KOCHVA 



BEHAVIOUR 




Fig. 1 Atractaspis engaddensis in a combined defensive/offensive 
position. Note the arched neck and the beginning of the coiled body 
with exposed tip of the tail. 

DISTRIBUTION 

The distribution of the Atractaspis species is unique (Fig. 2), starting 
from the Cape of South Africa, through the entire breadth of central 
Africa and along the Rift Valley to Arabia, Sinai, Jordan and Israel, 
reaching its northernmost border at Mount Gilboa ( Al-Oran & Amr, 
1995; Kochva, 1998). 

It is in Israel that the last species of Atractaspis was found and 
described. It was first recorded by Aharoni in 1945 as Atractaspis 
ate rri ma andlater described as anew species, Atractaspis engaddensis, 
by Haas in 1950. A. engaddensis is very similar to the Arabian A. 
microlepidota andersoni (Gasperetti, 1988), but a decision on the 
exact status of the two will have to await further information on the 
distribution of Atractaspis forms in the Arabian Peninsula as well as 
on the toxicity and composition of their venoms (see also Al-Sadoon 
etal, 1991; Al-Sadoon & Abdo, 1991; Schatti & Gasperetti, 1994). 



Species of the genus Atractaspis are desert-dwelling, fossorial 
snakes, whose behaviour and natural history are not well known. 
The Israeli species, A. engaddensis, is mostly found in the Negev 
desert and Dead Sea area, but it also extends to the Judean desert and 
along the Jordan Valley up to Mount Gilboa (Fig. 2). 

A. engaddensis feeds mainly on skinks, but also on lizards and 
geckoes that are caught at night above or below ground, beneath 
stones or other objects. In captivity, it also accepts baby mice and 
rats. A. microlepidota feeds on other snakes such Typhlops and 
Leptotyphlops, amphibians and small mammals, mostly rodents 
(Scortecci, 1939; Greene, 1997). In a four-year field study carried 
out by Akani etal. (2001) in south-eastern Nigeria, it was found that 
A. irregularis fed mainly on rodents, while A. aterrima and A. 
corpulenta ate lizards, skinks and snakes. 

The swallowing behaviour of Atractaspis may be influenced by 
its nearly vestigial teeth. As described for A. bibroni, it is character- 
ised by a rather inefficient transport mechanism in which the snake 
forces its head over the prey with lateral rotations around a vertical 
axis, rather than with the 'pterygoid walk' used by other snakes. 
This can be considered to be an adaptation towards feeding in 
narrow spaces and explained by the lack of connection between the 
pterygoid and palatine bones that are separated by a wide gap 
bridged by a ligament (Deufel & Cundall, 2000; MS; see also 
Underwood and Kochva, 1993). 

A. engaddensis lays 2-3 elongated eggs during the months of 
September-November and hatching occurs after about 3 months 
(Fig. 3). 

An interesting behavioural feature of this snake is its threat 
posture during which it presses its head to the ground while arching 
its neck (Fig. 1). This may turn either into a strike or into what 
appears to be a defensive display mechanism (Greene, 1979; 1997; 
Golani & Kochva, 1993). The snake forms a tight coil with the head 
hidden underneath the body and the wriggling tail is exposed above 




□ microlepidota group 

□ bibroni group 
ED overlapping area 



Fig. 2 Distribution map of Atractaspis species with the southern African bibroni group and the northern microlepidota group reaching the region of 
Mount Gilboa (after Underwood & Kochva, 1993; Joger, 1997). 



ATRACTASPIS - MINI REVIEW 

BMMKftflKf 



v * • 



J. 









5 



■> ' 




*"**w- 



* i* i 






* r 



* t 



Fig. 3 Hatching of Atractaspis engaddensis. 




Fig. 4 Defense mimicry posture by Atractaspis engaddensis, with hidden 
head and exposed tail. 




Fig. 5 Tail poking by Atractaspis engaddensis. 



93 

the coil so as to mimic the head (Fig. 4). The tail ends in a sharp tip 
that the otherwise immobile snake uses for poking when grasped at 
the posterior end of the body (Fig. 5). This behaviour may be 
mistaken for a genuine strike with the fang and deter any potential 
predator. Should that not suffice, there is always the hidden head that 
can be produced quickly from underneath the body coils and inflict 
a real, painful and dangerous strike. 



VENOM APPARATUS 

The venom apparatus of Atractaspis has not been dealt with in great 
detail beyond the general statement that the maxillary/fang unit is 
similar to that of vipers. This similarity is, however, superficial as 
the articulation between the prefrontal and maxilla in Atractaspis is 
in the form of a ball and socket articulation, which is more restricted 
in its movements, but apparently stronger (Pasqual. 1962). This 
condition may be important for the peculiar striking of these snakes, 
which is performed with one fang at a time while the mouth remains 
almost entirely closed (Fig. 6). Striking in this manner may be 
considered as a special adaptation for fossorial snakes that feed in 
narrow burrows underground. 

The venom fangs are relatively long and canaliculate and possess 
a blade-like ridge near the orifice of the fang (Kochva & Meier. 
1986), which may increase the wound and cause additional tissue 
damage during the strike, thus facilitating the spread of venom. 
Analyses of films taken during a strike through plastic sheathings 
show first the establishment of a firm contact of the head with the 
substrate, followed by the erection of the fang and piercing of the 
substrate by arching, lateral bending and downward rotation of the 
head (Fig. 6). Ejection of the venom is performed while the fang 
moves backward, further cutting through the surface (Golani & 
Kochva, 1988). 

The venom glands have a distinctive structure with secretion 
tubules arranged concentrically around the main lumen (Fig. 7). 
Unlike the viperids and elapids, there are no differentiated mucous 
accessory glands, but mucous cells are found in each of the secretion 
tubules close to the central lumen (Kochva et al., 1967). As in the 
other families of venomous snakes, there are species (the 
microlepidota group. Underwood & Kochva, 1993) with elongated 
venom glands that reach far beyond the corner of the mouth (Kochva, 
1959). The compressor muscle accompanies the gland along its 
entire length and probably squeezes it during the strike so as to 
increase the pressure in the central lumen and push the venom 
through the venom duct, fang canal and into the wound. The species 
with short glands (the bibrotti group) have a short, but thicker 
compressor. 

In a 756 mm long A. engaddensis the right gland reached the 30"' 
ventral and was 70 mm long, while in A. microlepidota it may reach 
one third of the body length - more than 300 mm in a specimen of 
900 mm (Scortecci, 1939). The left gland is usually longer than the 
right gland in both species and it is sometimes twisted along its 
longitudinal axis (Fig. 8). 



VENOM 

The venom of Atractaspis remained unknown for a long time, 
probably because not many serious bites were reported until now 
and it was thus ignored by toxinologists. In addition, the venom is 
very difficult to obtain not only because of the relative paucity of 
specimens collected, but also because of the difficulty in extracting 



94 



E. KOCHVA 




Fig. 6 Striking sequence by Atractaspis engaddensis; sequence from film 
(see text). 

it from the glands. Atractaspis, with its spade-shaped head, cannot 
be milked the way other venomous snakes are; a special method had 
to be devised as shown in Figure 9. 

Even today the biochemistry and pharmacology of the venom are 
known for only a few species, with almost all the information 
available originating from research with the venom of A. engaddensis. 

The toxicity of the venom varies among the species tested, the 
most potent venom being that of A. engaddensis, exceeding by 40 
times or more that of some other species (Table 1 ). It contains a set 







Fig. 7 Venom gland of Atractaspis engaddensis, cross section: M = 
compressor muscle, L = lumen of venom gland, T = secretion tubules. 

of enzymes not unlike those of other venomous snakes, a quite 
powerful hemorrhagin and a group of low-molecular weight toxins, 
the sarafotoxins, named after the Hebrew name of A. engaddensis - 
Saraf. When the venom was first fractionated by molecular sieving 
on a Sephadex G-50 column, 7 protein peaks were obtained. The 
first two contained high-molecular weight proteins with a 
hemorrhagic factor and the enzyme L-amino acid oxidase; the third 
peak showed phospholipase A, activity and peaks 5 and 6 contained 
very low molecular weight peptides, which made up 40% of the 
venom proteins and were highly toxic in i.v.-injected mice (Kochva 
et al.. 1982). These fractions were further purified resulting in 
several toxins characterised as sarafotoxins (SRTX), which showed 



Table 1 Toxicity of Atractaspis venoms, sarafotoxins. and mammalian 
endothelins in mice, LD (ng/g b.w.) 



A. bibroni 


500 


A. dahomeyensis 


2000 


A. microlepidota 


>2000 


A. micropholis 


>3000 


A. engaddensis 


75 


Sarafotoxin-a 


10 


Sarafotoxin-b 


10 


Sarafotoxin-c 


300 


Sarafotoxin-d/e 


>2000 


Endothelin-1 


15 


Endothelin-3 


30 






ATRACTASPIS - MINI REVIEW 



95 




Fig. 8 Elongated venom glands of Atractaspis engaddensis: V = venom gland. 




Fig. 9 Venom extraction from Atractaspis engaddensis using a piece of 
tubing for safety and a parafilm-covered lid for the collection of venom. 

a high degree of structural homology amongst themselves and with 
a group of active peptides that were isolated from mammalian 
endothelium, the endothelins (Fig. 10). The sarafotoxins and 
endothelins are also similar in their pharmacological activity and are 

15 10 



composed of 21 amino acid residues, with two disulphide bridges 
between Cys 1-15 and 3-1 1 (Yanagisawa et at., 1988; Takasaki et 
al., 1988; Wollberg et al.. 1990; Kochva et al, 1993). Another 
member of the sarafotoxin/endothelin family, bibrotoxin, was iso- 
lated from the venom of A. bibroni. It differs from SRTX-b in only 
one amino acid substitution and induces vasoconstriction in rat aorta 
(Becker et al.. 1993). The venom of A. m. microlepidota contains a 
series of peptides with a somewhat higher molecular weight that are 
composed of 24 amino acids (Ducancel et al, 1999) and are 
apparently less toxic than SRTX-a or b. 

The sarafotoxins and endothelins are now synthesised by pharma- 
ceutical companies and are widely used in both basic and applied 
research, both clinical and industrial, in the field of cardiology and 
in blood pressure studies (Ducancel et al., 1999: Yaakov et al., 
2000). 

The various sarafotoxins (and endothelins) differ in their activity 
and toxicity, the most potent ones being SRTX-a and SRTX-b 
(Table 1 ). which exert a strong influence on the cardiovascular 
system (Wollberg et al.. 1988). SRTX-b shows three, apparently 
separate, effects on the heart: 1 ) positive inotropicity, which is 
manifested by an increased contractility in isolated hearts and heart 
muscles and in in vivo injected mice with sublethal doses of the 

15 20 



Cys-Ser-Cys-Lys-Asp-Met-Thr-Asp-Lys-Glu-Cys-Leu-Asn-Phe-Cys-His-Gln-Asp-Val-Ile-Trp 



Cys 
Cys 
Cys 
Cys 



Cys * 
Cys * 
Cys * 
Cys -Ala 



Ser 



Ser 



Cys-Thr-Cys-Asn * * * 
Cys-Thr-Cys * * * * 
Cys * Cys-Asn * Ile-Asn 
Cys * Cys-Ser-Ser-Leu-Met 
Cys * Cys-Ser-Ser-Trp-Leu 
Cys * Cys-Asn-Ser-Trp-Leu 
Cys-Thr-Cys-Phe-Thr-Tyr- Lys 
Cys * Cys-Ala-Thr-Phe-Leu 
1 5 



Glu 



Tyr 

Tyr 

Tyr 

* 

Tyr 



Cys 

Cys 

Cys 

Cys 

Cys 

Cys 

Cys Met Tyr 

Cys-Val-Tyr 

Cys-Val-Tyr 

Cys-Val-Tyr 

Cys-Val-Tyr- 

Cys-Val-Tyr 



10 



* Cys 

* Cys 

* Cys 

* Cys 

* Cys 

* Cys 

* Cys 

* Cys 

* Cys 

* Cys 
Tyr-Cys 

* Cys 

15 



* Gly-Ile 

* * * 



Leu 
Leu 
Leu 
Leu 
Leu 



He 
He 
He 
He 
He 



SRTX-a 

* SRTX-al 

* SRTX-b 

* SRTX-bl 

* BTX 

* SRTX-c 

* SRTX-d/e 

* Asp-Glu-Pro A. microlepidota 

* ET-1 

* ET-2 

* VIC 

* ET-3 

* ET-trout (1999) 



20 



Fig. 10 Amino acid sequences of sarafotoxins and endothelins: BTX = bibrotoxin. ET = endothelin, SRTX = sarafotoxin. VIC = vasoactive intestinal 
contractor. 



96 



E. KOCHVA 



a ^, 



c ,• 




Jk^3 






. . lmuuuJ ' ' 1 1 — i — i 1 i 1 

M 0.5 1.0 sec. 




of other blood vessels and may be considered as one of the most 
potent vasoconstrictors. 

In human patients too, cardiotoxic symptoms and a rise in blood 
pressure were observed, but were considered as secondary develop- 
ments of some kind of neurotoxic effects. However, neither 
presynaptic nor postsynaptic neurotoxicity was observed in labora- 
tory tests with nerve-muscle preparations using whole A. engaddensis 
venom (Weiser et al, 1984); SRTX does show specific binding to 
different isolated regions of brain preparations, with the highest 
binding capacity found in the cerebellum, choroid plexus and hip- 
pocampus (Ambar et al, 1988), but its function there is not known. 

In human bites by A. engaddensis and A. irregularis, some of 
which were extremely severe, changes in the ECG were observed 
(Fig. 11), including S-T elevation or depression, flattening of the T- 
waves and prolonged P-R intervals pointing to myocardial ischemia 
and atrioventricular conduction abnormalities (Chajek et ai, 1974; 
Warrell, 1995; Kurnik, et al, 1999). The transient atrioventricular 
block that developed in a 17-year old boy bitten on his left foot was 
considered to be a secondary complication of the bite (Alkan and 
Sukenik, 1974), rather than a direct influence of the toxins on the 
heart. 

The other systemic symptoms, which may develop within min- 
utes, include fever, nausea, general weakness, sweating, pallor, 
fluctuations in the level of consciousness and a rise in blood pressure 
(Doucet & Lepesme, 1953; Chajek etai, 1974; Kurnik etal, 1999). 

Most bites were on the fingers and the local effects were demon- 
strated mainly by gross oedema of the hand that extended up to the 
forearm and shoulder (A. irregularis - Doucet and Lepesme, 1953; 
A. corpulenta - Gunders et al., 1960; A. microlepidota - Warrell et 
al, 1976) and by blistering and serous vesicles that appeared at the 
site of the bite and underwent hemorrhagic transformation (Fig. 12; 
Kurnik et al. 1999). In some previously reported cases (Chajek et 
al, 1974; Chajek & Gunders, 1977), local necrosis developed that 
required surgical intervention including amputation. In two cases, 
one by A. bibroni, the other by A. engaddensis, the bitten finger 
partially or fully recovered within a month, but tenderness of the 
bitten site remained for a long time (Stewart, 1965; Kurnik et al, 
1999). 

Although the bites by several species of Atractaspis, such as A. 
dahomeyensis, A. aterrima, A. corpulenta and A. bibroni were mild 
(Warrell et al, 1976; Tilbury & Branch, 1989), A. engaddensis, A. 
irregularis and perhaps other species should be regarded as dangerous 




H 



Fig. 11 ECG recording after Atractaspis engaddensis envenomation. M 
= mouse: v = venom injection; b - f : 120 - 600 seconds after venom 
injection. H = human: upper trace - at admission to the hospital: lower 
trace - 24 hours after the bite (see text). 

toxin; 2) direct effect of the toxin on the cardiac conducting system; 
and 3) cardiac ischemia, which is caused by constriction of the 
coronary blood vessels. The latter two cause severe A-V block, 
which may lead to cardiac arrest. The cardiotoxic effects are mani- 
fested by marked changes in the ECG, in both human victims and in 
mice injected with either SRTX-b or whole venom (Fig. 1 1 ). These 
changes include an increase in amplitude of the R- and T-waves, a 
prolongation of the P-R interval, 'dropped beats' and complete A- 
V block and cardiac arrest. In addition. SRTX-b causes contraction 




Fig. 12 Bitten index finger showing hemorrhagic transformation of 
serous vesicles. 



ATRACTASPIS - MINI REVIEW 



97 




Fig. 13 Ridges on the teeth of Pachyrhachis problematicus (arrow). 



new distribution record for engaddensis (Al-Sadoon & Abdo, 199 1 ; 
see also Al-Sadoon et al, 1991; Gasperetti, 19881; Joger, 1997; 
Schatti & Gasperetti, 1994). 

It should be pointed out that the toxicity of the venom of certain 
species, such as A. microlepidota, may vary according to distri- 
bution, causing death in certain cases (see above) or containing less 
potent toxins in others (above and Table 1). 

As with other venoms, snakes and some mammals are also 
resistant to Atractaspis engaddensis venom, including the local 
mongoose (Herpestes ichneumon, Bdolah et al, 1997). At least in 
one instance, it was found that a mongoose (Paracynictis selousi) 
fed on a specimen of A. bibroni (Greene, 1997). 

There is no antiserum available against any of the Atractaspis 
species. 



mainly because of their influence on the cardiovascular system, 
which may lead to death. Only a very small number of lethal cases 
has been recorded until now, perhaps a total of five, three by A. 
microlepidota (one adult man and two girls aged 4 and 6), one by A. 
irregularis, an adult man, and one unknown (Corkill et al., 1959; 
Warrell et al. , 1 976). Despite the fact that A. engaddensis has one of 
the most potent venoms known, all patients bitten by this species 
finally recovered, one probably due to 'the immediate and energetic 
treatment he received' (Chajek et al, 1974). Most recently (July, 
2002) a forty-six-year-old man was bitten on the inner aspect of the 
right thumb while trying to catch an Atractaspis engaddensis near 
his home in the Judean Desert, some 1 5 km north west of Jericho. He 
was taken to the hospital where he arrived after about 40 minutes in 
serious condition. Resuscitation failed and he was pronounced dead 
after about 45 minutes (Nadir & Stalnikowicz, personal communi- 
cation). This is the first death by an Atractaspis engaddensis bite in 
Israel. Another recent case, from Saudi Arabia, involved a two-year- 
old-girl who died within one hour after being bitten on the foot by 
what was identified as A. microlepidota engaddensis. The region 
where the bite occurred, at Diriyah near Riyadh, Saudi Arabia, is a 



EVOLUTION 



The discussion of snake origin and evolution has been recently 
revived by a new and renewed examination and analysis of the 
fossils discovered by Haas (1979; 1980a; 1980b) at an Upper 
Cretaceous site north of Jerusalem. While the debate on the eco- 
logical origin (marine or terrestrial) and the relationships of these 
specimens (mosasauroid or macrostomatan) is still going on (Lee 
& Caldwell, 1998; Greene & Cundall, 2000; Tchernov et al, 
2000), Rieppel & Zaher (2001) have recently concluded that 
'Pachyrhachis is neither a basal snake, nor a link between snakes 
and mosasauroids, but shows macrostomatan affinities instead'. 
Pachyrhachis possesses ridges or cutting edges on its teeth 
(Rieppel & Kearney, 2001; Fig. 13) and the teeth of another fossil, 
Haasiophis, have still to be further investigated in detail. Should 
furrows or any other suggestive structures be found, they could be 
taken as plausible signs for the existence of some kind of glands 
that might have secreted active substances, even before the 
appearance of caenophidian snakes. 



ATRACTASPIDIDAE 



VIPERIDAE 



SARAFOTOXINS HEMORRHAGES HEMORRHAGINS PHOSPHOLIPASE 

A 2 -CONTAINING TOXINS 
(VIPEROTOXINS; PRESYNAPTIC 
NEUROTOXINS) 




ELAPIDAE 

PRESYNAPTIC NEUROTOXINS 
POSTSYNAPTIC NEUROTOXINS 
CARDIOTOXINS; CYTOTOXINS 




ANCESTRAL 
ENDOTHELIN 



ANCESTRAL 
PROTEASE 



ANCESTRAL PHOSPHOLIPASE 



Fig. 14 Schematic representation of the possible origin of some major snake venom toxins from enzymatic precursors (partly after Strydom. 1979). 



98 



E. KOCHVA 



It has been suggested that a system that produced active sub- 
stances with the means of introducing them into the prey probably 
lay at the foundation of the major radiations of higher snakes 
(Underwood & Kochva, 1993). This system underwent further 
evolution in the Atractaspididae (mainly Atractaspis), Viperidae, 
Elapidae and several lineages of 'Colubridae'. 

Some of the active substances were probably enzymatic in nature 
and related to enzymes secreted by evolutionarily 'older' glands, 
such as the pancreas. Indeed, phospholipases found in the venom of 
Elapidae, for instance, show sequence homology with the enzymes 
secreted by the mammalian pancreas. Some of the ancestral en- 
zymes developed into toxins, such as hemorrhagins and neurotoxins, 
with or without loss of enzymatic activity (Fig. 14; Strydom, 1979; 
Kochva, 1987). 

Hemorrhagins are found in two families (Viperidae and 
Atractaspididae); presynaptic neurotoxins in two (Elapidae and 
Viperidae); and two families each possess a specific and unique 
group of toxins - postsynaptic neurotoxins in elapids and sarafotoxins 
in Atractaspis. 

The hemorrhagin found in the venom of Atractaspis is neutral- 
ised by antibodies against Vipera palaestinae venom (Ovadia, 
1987) and may thus be related to viperid hemorrhagins, originat- 
ing from some kind or kinds of protease. The presynaptic and 
postsynaptic neurotoxins, as well as the cytotoxins and 
cardiotoxins, apparently originate from phospholipase-like mol- 
ecules. The enzyme phospholipase A, may be part of the 
presynaptic neurotoxins and its enzymatic activity may still be 
essential for its toxicity. The postsynaptic neurotoxins, the 
cytotoxins and the cardiotoxins apparently underwent major 
changes including loss of enzymatic activity, chain shortening 
and gain of neurotoxicity (Strydom, 1979). 

The sarafotoxins are structurally very similar to the endothelins, 
which are evolutionarily highly conserved, and are found in all 
vertebrates, as well as in some invertebrate groups. It should be 
emphasised, however, that the genes of the mammalian endo- 
thelins were found on three separate chromosomes, whereas the 
sarafotoxin genes seem to be located on the same chromosome. 
The organisation of the SRTX genes of both A. engaddensis and 
A. in. microlepidota and their precursors are also different from 
those of the endothelins and may have evolved separately 
(Ducancel et ai, 1993; 1999). 

There is, of course, a great deal of information still missing, but 
the evolution of the sarafotoxins and of some of the other snake 
venom toxins and their use in feeding and defense may best be 
defined as exaptations; these are features that once had different 
functions but are now used in a new role that enhances the fitness of 
their bearers (Gould & Vrba, 1982). 



ACKNOWLEDGMENTS. I would like to thank first and foremost Garth 
Underwood for everything he taught me, and not only in Herpetology, and for 
a long and fruitful co-operation and warm friendship. I thank the editors of 
the Bulletin for inviting me to take part in this publication and Avner Bdolah, 
David Cundall, Alexandra Deufel, Dan Graur, Eyal Nadir, Olivier Rieppel, 
Ruth Stalnikowicz and Garth Underwood for comments on the manuscript 
and for sharing with me some of their unpublished findings. The remarks of 
the referees and the help of the Editor are also highly appreciated. 

I am very much indebted to my co-workers in Israel, South Africa and 
Japan for their major share in the different disciplines of the Atractaspis 
research and to the undergraduate and graduate students (listed in the 
references) and to many more who picked up the sarafotoxin project (with or 
without endothelin) and developed it into such a broad and deeply interesting 
field, with still much more to be expected in the future. 



Special thanks are due to Moshe Alexandroni, Lydia Maltz, Omer 
Markowitz, Naomi Paz, Amikam Shoob and Varda Wexler for help with the 
illustrations and with the preparation of the manuscript. 



REFERENCES 



Aharoni, I. 1945. Animals hitherto unknown to or little known from Palestine. Bulletin 

of the Zoological Society of Egypt, Supplement (6): 40—11. 
Alkan, M.L. & Sukenik, S. 1974. Atrioventricular block in a case of snakebite 

inflicted by Atractaspis engaddensis. Transactions of the Royal Society of Tropical 

Medicine and Hygiene 69: 166. 
Akani, G.C., Luiselli, L.M., Angelici, F.M., Corti, C. & Zuffi, M.A.L. 200 1 . The case 

of rainforest stiletto snakes (genus Atractaspis) in southern Nigeria. Evidence of 

diverging foraging strategies in grossly sympatric snakes with homogeneous body 

architecture? Ethology, Ecology and Evolution 13: 89-94. 
Al-Oran, R.M. & Amr, Z.S. 1995. First record of the Mole Viper. Atractaspis micro- 
lepidota engaddensis, from Jordan. Zoology in the Middle East. Reptilia 11: 47^49. 
Al-Sadoon, M.K. & Abdo, N.M. 1991. Fatal envenoming by the snake Atractaspis 

newly recorded in the central region of Saudi Arabia. Journal of King Saud 

University, Science 3: 123-131. 
, Al-Farraj, S.A. & Abdo, N.M. 1 99 1 . Survey of the reptilian fauna of the Kingdom 

of Saudi Arabia. III. An ecological survey of the lizard, amphisbaenian and snake 

fauna of Al-Zulfi area. Bulletin of the Maryland Herpelological Society 27: 1-22. 
Ambar, I., Kloog, Y., Kochva, E., Wollberg, Z., Bdolah, A., Oron, U. & Sokolovsky, 

M. 1988. Characterization and localization of a novel neuroreceptor for the peptide 

sara-fotoxin. Biochemical and Biophysical Research Communications 157: 1 104 — 

11 10. 
Bdolah, A., Kochva, E., Ovadia, M., Kinamon, S. & Wollberg, Z. 1997. Resistance 

of the Egyptian mongoose to sarafotoxins. Toxicon 35: 1251-1261. 
Becker, A., Dowdle, E.B., Hechler, U., Kauser, K., Donner. P. & Schleuning, WD. 

1993. Bibrotoxin. a novel member of the endothelin/sarafotoxin peptide family, from 

the venom of the burrowing asp Atractaspis bibroni. Federation of European 

Biochemical Societies (FEBS) 315: 100-103. 
Bourgeois, M. 1 96 1 . Atractaspis - a misfit among the Viperidae? News Bulletin of the 

Zoological Society of South Africa 3: 29. 

1968. Contribution a la morphologic comparee du crane des ophidiens de 

VAfrique centrale. Vol. XVIII. Publications de l'Universite Officielle du Congo a 
Lubumbashi: 1-293. 

Chajek, T., Rubinger, D., Alkan, M., Melamed, R.M. & Gunders, A.E. 1974 

Anaphylactoid reaction and tissue damage following bite by Atractaspis engaddensis. 

Transactions of the Royal Society of Tropical Medicine and Hygiene 68: 333-337. 
& Gunders, A.E. 1977. Clinical and biochemical observations following 

bites of Atractaspis engaddensis. Rofe Hamishpaha 7: 119-122. 
Corkill, N.L., Ionides, C.J. P. & Pitman, C.R.S. 1959. Biting and poisoning by the 

mole vipers of the genus Atractaspis. Transactions of the Royal Society of Tropical 

Medicine and Hygiene 53: 95-101 . 
Deufel, A. & Cundall, D. 2000. Feeding in stiletto snakes. American Zoologist 40: 

996-997. 

Feeding of Atractaspis (Serpentes: Attractaspididae): A study in conflicting 

functional constraints. (MS) 

Ducancel, F., Marte, V., Dupont, C, Lajeunesse, E., Wollberg, Z., Bdolah, A., 

Kochva, E., Boulain, J.-C. & Menez, A. 1993. Cloning and sequence analysis of 

cDNA encoding precursors of sarafotoxins. Journal of Biological Chemistry 268: 

3052-3056. 
, Wery, M., Hayashi, M.A.F., Muller, B.H., Stocklin, R. & Menez, A. 

1 999. Les sarafotoxines de venins de serpent. Annates de I Tnstitut Pasleur/Actualites 

10: 183-194. 
Doucet, J. & Lepesme, P. 1953. Sur un cas d'envenimation par Atractaspis, viperide 

ouest-africain. Bulletin d'/nstitut francais d'Afrique noire 15: 855-859. 
Gasperetti, J. 1988. Snakes of Arabia. Fauna of Saudi Arabia 9: 169^150. 
Golani, I. & Kochva E. 1988. Striking and other offensive and defensive behaviour 

patterns in Atractaspis engaddensis (Ophidia. Atractaspididae). Copeia 1988: 792- 

797. 

1993. Tail display in Atractaspis engaddensis (Atractaspididae. Serpentes). 

Copeia 1993: 226-228. 

Gould, S.J. & Verba, E.S. 1982. Exaptation - a missing term in the science of form. 

Paleobiology 8: 4-15. 
Gravlund, P. 2001. Radiation within the advanced snakes (Caenophidia) with special 

emphasis on African opistoglyph colubrids. based on mitochondrial sequence data. 

Biological Journal of the Linnean Society 72: 99-114. 
Greene, H.W. 1979. Behavioral convergence in the defensive displays of snakes. 

Experientia 35: 747-748. 
1997. Snakes, the Evolution of Mystery in Nature. University of California Press. 

Berkeley. Los Angeles. London, pp. 35 1 . 



ATRACTASPIS- MINI REVIEW 



99 



& Cundall, D. 2000. Limbless tetrapods and snakes with legs. Science (287): 

1939-1941. 
Gunders, A.E., Walter, H.J. & Etzel, E. 1960. Case of snake-bite by Atractaspis 

corpulenta. Transactions of the Royal Society of Tropical Medicine and Hygiene 54: 

279-280. 
Gunther, A. 1858. Catalogue of the cohtbrine snakes in the collection of the British 

Museum. British Museum (Natural History). London, 281 pp. 
Haas, G. 193 1 . Uber die Morphologie der Kiefermuskulatur und die Schadelmechanik 

einiger Schlangen. Zoologisch.es Jahrhuch, Abteilung anatomie 54: 333—116. 

1950. A new Atractaspis (Mole Viper) from Palestine 1950: 52-53. 

1979. On anew snakelike reptile from the Lower Cenomanian of Ein Jabrud, near 

Jerusalem. Bulletin du Museum national d'Histoire naturelle. Section C: Sciences de 

la Terre. Serie 4 (1): 51-64. 

1980a. Pachyrhachis prohlematicus Haas, snakelike reptile from the lower 

Cenomanian: ventral view of the skull. Bulletin du Museum national d'Histoire 
naturelle, Section C: Sciences de la Terre. Serie 4 (2): 87-104. 

1 980b. Remarks on a new ophiomorph reptile from the lower Cenomanian of Ein 

Jabrud. Israel, pp. 177-192. In L.L. Jacobs (ed.). Aspects of Vertebrate History- 
essays in honor of Edwin Harris Colbert. Museum of Northern Arizona Press, 
Flagstaff, 393 pp. 

Joger, U. 1997. Tubinger Atlas des Vorderen Orients, Giftschlangen (Middle East 
Venomous Snakes). Dr. Ludwig Reichert Verlag, Wiesbaden. 

Kochva, E. 1959. An extended venom gland in the Israel Mole Viper. Atractaspis 
engaddensis Haas 1950. Bulletin of the Research Council of Israel B8: 31-34. 

1987. The origin of snakes and evolution of the venom apparatus. Toxicon 25: 65- 

106. 

1998. Venomous snakes of Israel: Ecology and snakebite. Proceedings of the joint 

Israeli-Jordanian Conference on Venomous Animal Bites. Amman. Jordan. I July 
1998. Ofer Shpilberg & Dani Cohen. Eds. Public Health Reviews 26: 209-232. 

, Bdolah, A. & Wollberg, Z. 1993. Sarafotoxins and endothelins: evolution. 

structure and function. Toxicon 31: 541-568. 

& Meier, J. 1986. The fangs of Atractaspis engaddensis Haas (Serpentes: 

Atractaspididae). Revue Suisse de Zoologie 93: 749-754. 

, Shayer- Wollberg, M. & Sobul, R. 1967. The special pattern of the venom gland 

in Atractaspis and its bearing on the taxonomic status of the genus. Copeia 1967: 

763-772. 
, Viljoen, C.C. & Botes, D.P. 1982. A new type of toxin in the venom of snakes 

from the genus Atractaspis (Atractaspidinae). Toxicon 20: 581-592. 
Kurnik, D., Haviv, Y. & Kochva, E. 1999. A Snake bite by the burrowing asp. 

Atractaspis engaddensis. Toxicon 37: 223-227. 
Lee, M.S.Y. & Caldwell, M.W. 1998. Anatomy and relationships of Pachyrhachis 

prohlematicus, a primitive snake with hindlimbs. Philosophical Transactions of the 

Royal Society. London. B 352: 1521-1552. 
Ovadia, M. 1987. Isolation and characterization of a hemorrhagic factor from the venom 

of the snake Atractaspis engaddensis (Atractaspididae). Toxicon 25: 621-630. 
Pasqual, J.R.H. 1962. The biting mechanism of Atractaspis. The Nigerian Held 27: 

137-141. 
Reinhardt, I.T. 1843. Beskrivelse of nogle nye Slangearter. Kongelige danske 

Videnskabemes Selskabs. Afhandlinger (10): 233 -279. 
Rieppel, O. & Kearney, M. 2001. The origin of snakes: limits of a scientific debate. 

Biologist (48): 110-114. 



& Zaher, H. 2001 . Re-building the bridge between mosasaurs and snakes. Neues 

Jahrhuch fiir Geologic und Paldontologie. Abhandlungen (221): 1 1 1-132. 
Schatti, B. & Gasperetti, J. 1994. A contribution to the herpetofauna of Southwest 

Arabia. Fauna of Saudi Arabia 14: 348^119. 
Scortecci, G. 1939. Gli Oftdi Velenosi dell' Africa Italiana. Istituto Sieroterapico 

Milanese, Milan, 287pp. 
Smith, A. 1848. Illustrations of the Zoology of South Africa. Reptilia. Smith, Elder & 

Co.. London. 
Stewart, M.M. 1965. A bite by a burrowing adder. Atractaspis bihronii. The Journal 

of the Herpetological Association of Rhodesia (23/24): 47-50. 
Strydom, D.J. 1979. The evolution of toxins found in snake venoms. In: Snake 

Venoms. Handbook of Experimental Pharmacology. (C.Y. Lee. ed.) (52): 258-275. 

Springer Verlag, Berlin, 1 130 pp. 
Takasaki, C, Tamiya, N., Bdolah, A., Wollberg, Z. & Kochva, E. 1988. Sarafotoxin 

S6: Several isotoxins from Atractaspis (Burrowing Asps) venom that affect the 

heart. Toxicon (26): 543-548. 
Tchernov, E., Rieppel, O., Zaher, H., Polcyn, M.J. & Jacobs, L.L. 2000. A fossil 

snake with limbs. Science (287): 2010-2012. 
Theodor, J. & Albeck, C. (eds) 1966. Midrash Bereshil Raba, (5, 7), Jerusalem. 
Tilbury, C.R. & Branch, W.R. 1989. Observations on the bile of the southern 

burrowing asp (Atractaspis bihronii) in Natal. South African Medical Journal (75): 

327-331. 
Underwood, G. 1967. A Contribution to the Classification oj Snakes. British Museum 

(Natural History) Publication No. 653. Trustees of the British Museum (Natural 

History), London. 179pp. 
& Kochva, E. 1993. On the affinities of the burrowing asps Atractaspis (Serpentes: 

Atractaspididae). Zoological Journal of the Linnean Society 107: 3-64. 
Warrell, D.A. 1995. Clinical toxicology of snakebite in Africa and Middle East/ 

Arabian Peninsula, pp. 433-492. In: Clinical Toxicology of Animal Venoms and 

Poisons (Meier J. & J. White, eds.). CRS Press, Boca Raton, Florida, 752 pp. 
, Ormerod, L.D. & Davidson, N. McD. 1976. Bites by the night adder (Causus 

maculatus) and burrowing vipers (genus Atractaspis) m Nigeria. The American 

Journal of Tropical Medicine and Hygiene 25: 517-524 
Weiser, E., Wollberg, Z., Kochva, E. & Lee, M.S.Y. 1 984. Cardiotoxic effects of the 

venom of the burrowing asp. Atractaspis engaddensis (Atractaspididae. Ophidia). 

Toxicon 22: 767-774. 
Wollberg, Z., Bdolah, A. & Kochva, E. 1990. Cardiovascular effects of mammalian 

endothelins and snake venom sarafotoxin. In: Calcium Channel Modulators in Heart 

and Smooth Muscle: Basic Mechanisms and Pharmacological Aspects, pp. 283-299 

(Abraham, S. & G. Amilai, eds.). VCH. Weinheim/Deerfield Beach, Florida and 

Balaban, Rehovot/Philadelphia. 
— , Shabo Shina, R., Intrator, N., Bdolah, A., Kochva, E., Shavit, G, Oron, Y., 

Vidne, B.A. & Gitter, S. 1988. A novel cardiotoxic polypeptide from the venom of 

Atractaspis engaddensis (Burrowing Asp): Cardiac effects in mice and isolated rat 

and human heart preparations. Toxicon 26: 525-534. 
Yaakov, N.. Bdolah, A., Wollberg, Z., Ben-Haim, S.A. & Oron, U. 2000. Recovery 

from sarafotoxin-b induced cardiopathologicaJ effects in mice following low energy 

laser irradiation. Basic Research in Cardiology 95: 385-389. 
Yanagisawa, ML, Kurihara, H., Kimura, S., Tomobo, Y., Kobayashi, ML, Mitsui, Y., 

Yazaki, Y., Goto, K. & Masaki, T. 1988. A novel potent vasoconstrictor peptide 

produced by vascular endothelial cells. Nature 332: 41 1—415. 



I 



KX( ^57^6 (. n 



Bull. not. Hist. Mus. hand. (Zool.) 68(2): 101-106 



Issued 28 November 2002 



Origin and phylogenetic position of the Lesser 
Antillean species of Bothrops (Serpentes, 
Viperidae): biogeographical and medical 
implications 

WOLFGANG WUSTER AND ROGER S. THORPE 

School of Biological Sciences, University of Wales, Bangor LL57 2UW, UK 

MARIA DA GRACA SALOMAO 

Laboratorio de Herpetologia, Instituto Butantan, Av. Vital Brazil 1500, 05503-900 Sao Paulo-SP, Brazil 

LAURENT THOMAS 

Service des Urgences, CHRU, 97200 Fort de France, Martinique (French West Indies) 

GIUSEPPE PUORTO 

Museu Biologico, Instituto Butantan, Av. Vital Brazil 1500, 05503-900 Sao Paulo-SP, Brazil 

R. DAVID G. THEAKSTON 

Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UK 

DAVID A. WARRELL 

Centre for Tropical Medicine, University of Oxford, John Radcliffe Hospital, Headington, Oxford 0X3 9DU, 
UK 

SYNOPSIS. We use mitochondrial DNA sequences to infer the origin and phylogenetic position of the Lesser Antillean species 
of the pitviper genus Bothrops, B. carihbaeus and B. lanceolatus. The two species form a monophyletic group, which in turn forms 
the sister clade to the Bothrops asper-atrox complex. High levels of sequence divergence among the Caribbean species, and 
between them and the nearest mainland relatives, suggest a relatively ancient origin of these snakes. The hypothesis that the 
Lesser Antillean Bothrops are the result of a recent colonisation event from within the South American B. atrox complex is 
rejected, as is the hypothesis that they were introduced to their island habitats by aboriginal humans. The high level of 
morphological apomorphy displayed by B. lanceolatus suggests a stepping-stone colonisation, St. Lucia being colonised first and 
then Martinique from St. Lucia. The medical implications of these findings are discussed: a recent case of envenoming from Saint 
Lucia suggests that Bothrops caribbaeus causes the same thrombotic syndrome of envenoming as B. lanceolatus. 



INTRODUCTION 



The genus Bothrops Wagler, 1824 contains most of the pitviper 
fauna of South America. The genus (including the arboreal species 
sometimes assigned to Bothriopsis) contains approximately 36 
species, with a wider variety of body shapes and natural history traits 
than in any other New World pitviper genus. This greater diversity 
has been ascribed to the fact that Bothrops was the first group of 
pitvipers to reach the South American continent, thus giving ample 
opportunity for adaptive radiation (Wiister et al, in press). 

Two species of Bothrops occur in the Lesser Antilles: Bothrops 
caribbaeus (Garman, 1887) on St. Lucia, and Bothrops lanceolatus 
(Lacepede, 1789) on Martinique. The status and origin of these 
forms has been the subject of much debate. Long considered to be 
conspecific with Bothrops atrox, B. lanceolatus was revalidated by 
Hoge (1952), and the validity of B. lanceolatus and B. caribbaeus 
confirmed by Lazell (1964). This latter interpretation has been 
followed by most authors since then (e.g., Campbell & Lamar, 
1 989). However, Sandner Montilla (1981,1 990) regarded the Lesser 
Antillean Bothrops as conspecific with each other, as well as with 
mainland Bothrops asper and the northern Venezuelan populations 
of the B. asper-atrox complex. 

The origin of the Antillean Bothrops has been the subject of much 
speculation and mythology. This includes popular tales that the 



snakes were originally introduced by Carib Indians in their attempts 
to gain control of the islands from resident Arawaks (Dowling, 
1965), and the notion that dispersal from the South American 
mainland is common and ongoing (Sandner Montilla, 1981). 

The reptile fauna of the Lesser Antilles is primarily the result of 
long-distance dispersal by individual species, as these islands have 
not been linked to the South American mainland or any other 
landmass at any time in their history (Thorpe et al, in press; 
Malhotra and Thorpe 1 999). This means that some species present in 
these islands represent long-standing endemic lineages (Thorpe et 
al., in press; Malhotra and Thorpe 2000), whereas others appear to 
be the result of relatively recent dispersal events from well-defined 
source populations or taxa in South America, as is the case for the 
genus Corallus (Henderson & Hedges, 1995). 

Compared to morphological data, molecular markers such as 
mitochondrial DNA (mtDNA) sequence data have the advantage 
that they can give an estimate of phylogeny reasonably free of the 
confounding effects of differing natural selection pressures on the 
external phenotype. Moreover, molecular sequence data also have 
the advantage that they can give at least an approximate estimate of 
times of divergence between lineages, although the interpretation of 
molecular clocks is subject to various analytical and empirical 
problems (Hillis et al, 1996). 

Several recent mtDNA-based phylogenetic analyses of the genus 
Bothrops have included the Antillean species. Salomao et al. (1997, 



) The Natural History Museum, 2002 



102 



W. WUSTER£T/U. 





PLATE 1 

Bothrops caribbeus © R.S. Thorpe 

1999), using 580 b.p. of cytochrome b sequence, found B. caribbaeus 
and B. lanceolatus to the sister species of the South American 
populations of the B. atrox complex. However, the study included 
only a limited sampling of South American members of the B. atrox 
complex, and did not include representatives of B. asper from 
Central America. 

The aim of this paper is to explore in more depth the origin of the 
Antillean Bothrops, and its implications for other fields, using an 
expanded dataset of more sequence information from a larger number 
of potentially related species. 



MATERIALS AND METHODS 

We obtained tissue (ventral scale clippings or tail tips) and/or blood 
samples from species representing the principal clades within the 
genus Bothrops (including Bothriopsis), as well as the closely 
related Bothrocophias - see Wiister et al. (in press). We also 
included samples of the B. asper-atrox complex from around the 
coast of South and Central America, as these have been considered 
to be potential founder populations from which the ancestor of 
Antillean Bothrops could have arisen. For outgroup rooting, we used 
sequences of Bothrops alternatus and Bothrocophias microphthal- 
mias. Two regions of the mitochondrial DNA molecule were amplified 
using the polymerase chain reaction (PCR): a 767 base pair (bp) 
section of the gene for cytochrome b (cyib), and a 900 bp region of 
the gene for NADH dehydrogenase subunit 4 (ND4). Details of 
primers and laboratory protocols are given in Pook et al. (2000). 

Sequences were aligned by eye against published Bothrops 
sequences (Puorto et al., 2001). In order to test for the presence of 
saturation of certain categories of substitution, we calculated Tamura- 
Nei distances between all samples. This takes into account deviations 
from equal base compositions and differences in substitution rates 
among nucleotides. We then plotted unadjusted p-distances for 
transitions and transversions, and for the three codon positions 
separately, against Tamura-Nei distances. A decline in the rate of 
accumulation of individual categories of substitution with increased 
Tamura-Nei distances indicates saturation of that substitution category. 

We checked all sequences for insertions, deletions or the presence 
of stop codons. Any of these would have indicated that the sequences 
represent nuclear insertions of the mitochondrial genes (Zhang and 
Hewitt, 1996). The sequence data were assayed for the presence of 
a significant phylogenetic signal by means of the gl tree skewness 



Bothrops lanceolatus © D. Warrell 

statistic (Hillis and Huelsenbeck, 1992), calculated from 100,000 
trees randomly generated by PAUP* 4.0b8 (Swofford, 2001). 
Sequence divergences between clades were estimated using the 
program Phyltest (Kumar, 1996). 

We analysed our sequence data using both maximum parsimony 
(MP) and maximum likelihood (ML) as optimality criteria. Using 
multiple optimality criteria should identify those parts of a 
phylogenetic tree that are supported independently of the optimality 
criterion used. Such nodes should inspire greater confidence than 
nodes that are unstable and vary depending on method of analysis. 
All analyses were carried out using the program PAUP* 4.0b8 
(Swofford, 2001). 

For MP analyses, we selected Bothrops alternatus and B. micro- 
phthalmia as outgroups. We employed the heuristic search algorithm 
of PAUP* 4.0b8, using TBR branch swapping and 100 random 
addition sequence replicates. The analysis was carried out on the 
unweighted data only. 

The extent to which individual nodes on the tree were supported 
by the data was assessed using bootstrapping and Bremer (1994) 
branch support. Non-parametric bootstrap was implemented using 
heuristic searching, 1000 replicates, TBR branch swapping and 10 
random-addition-sequence replicates per bootstrap replicate. Bremer 
branch support for individual nodes was calculated through the use 
of the converse constraint option of PAUP*. 

For ML analyses, we identified the most appropriate model of 
sequence evolution through the use of the MODELTEST software 
(Posada & Crandall, 1998). A first ML search was run, using 
heuristic searching, a neighbour-joining starting tree and TBR branch 
swapping, and the sequence evolution parameters identified by the 
Modeltest software. These parameters were then re-estimated from 
the resulting ML tree, and a further search run using these re- 
estimated parameters. This was repeated until further estimates 
yielded no further changes of parameter values or tree likelihood 
scores. Bootstrap analysis involved 100 replicates, using NJ starting 
trees and NNI branch swapping. 

An important consideration of any proposed scientific hypothesis 
is whether the data supporting it can reject alternative hypotheses 
with statistical significance. In other words, do the data allow us to 
reject the null hypothesis that differences in tree optimality between 
the optimal tree and trees consistent with alternative hypotheses are 
due to random chance? In the case of the Antillean Bothrops, we 
tested the following alternative phylogenetic hypotheses: (i) non- 
monophyly of the Antillean Bothrops, i.e., the Antillean populations 
of Bothrops result from separate colonisation events; (ii) non- 



ANTILLEAN BOTHROPS 



103 



73/6 



iQooa 



microphthalmus 

alternatus 

jararaca 



insularis 



81/8 



52/3 



taeniatus 



96/12 



pulcher 
— punctatus 



56/2 



100/19 



100/14 



24ZZ 



jararacussu 

brazili 
lanceolatus 



84/5 



100/18r caribbaeus 
L caribbaeus 

^^ — ~ asper - Belize 

1Q0/gf* atrox - Suriname 
"Tr a/rox - Fr. Guyana 

94 |^89/2 afrox " Gu y ana 

leucurus - D. Martins 
leucurus - Salvador 

86/2 1 - leucurus - P. Seguro/T. Freitas 
39/6_ 

f colombiensis - S. Francisco 



100/15 



100/13l|'e 



10 



n.s./1 
65/1 




colombiensis - Guaibacoa 
colombiensis - A.d.Orituco 
afrox - Sta. Izabel 
marajoensis - I. Marajo 
50/1U — atrox - S. Bento 
61 /1* marajoensis - S. C. Arari 

microphthalmus 
alternatus 



80 



100 



67 



81 



jararaca 
insularis 
— taeniatus 



pulcher 



99 



100 



k2 



100 



100 



jararacussu 

brazili 

— punctatus 
lanceolatus 



100 



{. 



87 




0.1 



caribbaeus 
caribbaeus 

asper - Belize 
colombiensis - A. d. Orituco 
atrox - Sta. Izabel 
marajoensis - 1. Marajo 
atrox - S. Bento 
marajoensis - S. C. Arari 

10Q colombiensis - S. Francisco 

' — colombiensis - Guaibacoa 
afrox - Suriname 
afrox - Fr. Guyana 
97 Q4 1 - afrox - Guyana 

100 T leucurus ' D. Martins 
|| leucurus - Salvador 
88l- leucurus - P. Seguro/T. Freitas 



100f £ 
)7dX 



Fig. 1 Maximum parsimony (top) and maximum likelihood (bottom) estimates of the phylogeny of Both rops. In the MP tree, numbers before the slash 
refer to bootstrap support, numbers after the slash indicate Bremer support. In the ML tree, numbers on nodes indicate bootstrap support. 



104 



W. WUSTER ETAL. 



monophyly of the B. asper-atrox complex, i.e., that the Antillean 
populations originate from within the B. asper-atrox complex; (iii) 
non-monophyly of the South American B. atrox complex, i.e., that 
the Antillean species originate from within the cis-Andean B. atrox 
complex, paralleling the phylogeography of Corallus (Henderson & 
Hedges, 1995); and (iv) monophyly of B. caribbaeus, B. lanceolatus, 
B. asper and northern Venezuelan populations of the B. asper-atrox 
complex to the exclusion of the cis-Andean B. atrox complex, as 
implied by the classification of Sandner Montilla (1990). We used 
Wilcoxon signed-ranks (WSR) tests (Templeton tests - Templeton, 
1983) to compare the optimal MP tree and MP trees depicting 
alternative hypotheses, and Shimodaira-Hasegawa (SH) tests 
(Shimodaira & Hasegawa, 1999) to compare the corresponding ML 
trees. 



RESULTS 

We aligned a total of 1401 b.p. of mtDNA sequence information 
(ND4: 693 b.p.; cytfr: 708 b.p.). The sequences included no indels or 
stop or other nonsense codons, and contained the usual bias towards 
transitions and substitutions concentrated into third codon positions 
typical of mitochondrial DNA. We conclude that our sequences 
represent mtDNA rather than nuclear insertions. Samples are listed 
in Appendix 1. The 100,000 random trees generated a skewness 
statistic of gl=-0.599403, rejecting the null hypothesis that the data 
contain no significant phylogenetic signal (P < 0.01; Hillis and 
Huelsenbeck, 1992). 

Levels of sequence divergence among the taxa included ranged 
from 0.3% to 13.65% (unadjusted p-distance). Bothrops caribbaeus 
and B. lanceolatus differ from each other by 4.3%, and from the B. 
asper-atrox group by an average of 5.77% and 6. 15% respectively, 
with an average divergence of 5.9% when the Antillean haplotypes 
are treated as a single clade. Levels of sequence divergence within 
the B. asper-atrox clade range from 0.3% to 5.5% 

The MP analysis resulted in a single most parsimonious tree of 
1030 steps (CI 0.5398; HI 0.4602; RI 0.6465). In this tree, the two 
Antillean taxa formed a clade, which in turn forms the sister clade of 
all samples of the Bothrops asper-atrox complex (Fig. 1 ). 

The MODELTEST software identified the GTR+I+G model, a 
submodel of the general time-reversible model (Yang etal., 1994) as 
optimal for the data at hand. A ML tree was constructed using the 
parameters calculated by MODELTEST, and the parameters were 
recalculated from the resulting tree, and used in a further ML search, 
which resulted in a tree with the likelihood score -ln(L)= 6652.69 1 22. 

Further estimates of sequence evolution parameters did not result 
in any change of parameter values or tree likelihood score (Fig. 1). 
The MP and ML trees differ only in branching order within the cis- 
Andean B. atrox complex, and in the relative position of the B. 
jararacussu-brazili clade and B. punctatus. 

The results of our tests of alternative tree topologies are shown in 



Table 1. Neither the WSR nor the SH test significantly reject the 
possibility that the two Antillean species may be the result of 
separate colonizations of the Lesser Antilles, although the result of 
the SH test was almost significant. They do, however, significantly 
reject the hypothesis that the Antillean species originate from within 
the cis-Andean radiation of the B. asper-atrox complex, and also 
reject Sandner Montilla's suggestion of conspecificity between B. 
lanceolatus, B. caribbaeus, B. asper and northern Venezuelan 
Bothrops, to the exclusion of other South American populations of 
the B. atrox group. 



DISCUSSION 

Our results confirm the position of the Antillean species of Bothrops 
as the sister clade of the Bothrops asper-atrox complex, as suggested 
by Salomao ef a/. (1997, 1999) and Wustere/a/. (1997, 1999). The 
monophyly of the Antillean taxa is supported by high bootstrap and 
Bremer support values, although a tree supporting this arrangement 
is not significantly longer than the optimal tree constrained not to 
include this clade. 

The high level of sequence divergence between the Antillean 
Bothrops and their mainland relatives (5.9%) is consistent with a 
lineage split dating back to the Miocene or earliest Pliocene. Wiister 
et al. (in press) suggested a rate of sequence evolution for cytfr and 
ND4 of between 0.66 and 1.4% My -1 in Neotropical pitvipers. This 
would date the timing of the split between the Antillean Bothrops 
clade and the B. asper-atrox clade at 4.2-8.9 Mya, i.e., the late 
Miocene or earliest Pliocene. Similarly, the split between B. 
caribbaeus and B. lanceolatus (sequence divergence: 4.3%) can be 
dated to 3.1-6.5 Mya. Hedges (1996) estimated the divergence of 
the B. asper-atrox complex to have taken place within the last 4 My, 
and assumed dispersal to the Antilles to have taken place during that 
timeframe, whereas our data suggest a slightly earlier lineage split. 
In any case, it can be concluded that the two Antillean Bothrops 
species represent two relatively old. independent lineages. Obviously, 
in view of the errors inherent in any attempt at molecular clock 
usage, these estimates should be treated as approximations rather 
than exact timings. 

The notion that these populations are the result of a recent 
dispersal event from within South America, as is the case in West 
Indian Corallus (Henderson & Hedges, 1995), is refuted by both 
tree topology and statistical tree comparison tests. Equally, the 
notion that the presence of these snakes in the Lesser Antilles is the 
result of a primitive form of biological warfare among aboriginal 
people (Dowling, 1965) will have to be abandoned, despite its 
romantic appeal. 

The colonisation sequence of the two species can be resolved 
from morphological data, particularly scalation. In terms of dorsal 
and ventral scale counts, B. caribbaeus is indistinguishable from 
many populations of the B. asper-atrox complex. On the other hand. 



Table 1 Differences in tree length or likelihood, statistics, and their significance, between the most parsimonious or the most likely trees, and trees 
constrained to be compatible with alternative phylogenetic or biogeographical hypotheses. 



d(steps) 



Wilcoxon signed-ranks 
- z 



Shimodaira-Hasegawa 
d(lnL) P 



Non-monophyly of B. caribbaeus and B. lanceolatus 
Non-monophyly of B. asper-atrox complex 
Non-monophyly of cis-Andean B. atrox complex 
Monophyly of B. caribbaeus, lanceolatus, asper 
and northern Venezuelan populations 



7 


1.4000 


0.1615 


15.05075 


0.054 


5 


1.1471 - 1.5076 


0.1317-0.2513 


3.15866 


0.181 


5 


2.4019 


0.0163* 


20.06423 


0.018* 


8 


2.9200-3.0870 


0.002-0.0035* 


24.34213 


0.005* 






ANTILLEAN BOTHROPS 



105 



B. lanceolatus has higher ventral and dorsal scale row counts than 
practically all populations of the B. jararacussu-punctatus-atrox 
clade. This suggests that the extreme scale counts found in B. 
lanceolatus represent an autapomorphy compared to B. caribbaeus 
and mainland Bothrops. This makes a hypothesis of dispersal from 
the mainland to St. Lucia, and then a further dispersal event to 
Martinique, more parsimonious than dispersal to Martinique 
followed by further dispersal to St. Lucia. Since St. Lucia lies 
between South America and Martinique, this scenario is also more 
geographically parsimonious than the alternative. The slightly 
greater length of the branch leading to B. lanceolatus is also consist- 
ent with this hypothesis (De Salle & Templeton, 1 988; Thorpe etai, 
1994). 

An understanding of the phylogenetic position of Bothrops 
caribbaeus and B. lanceolatus may also have implications for their 
venom composition and the treatment of snakebite in the Caribbean. 
Bothrops lanceolatus envenoming has been documented to produce 
a unique syndrome different from that of other species of Bothrops. 
In addition to local symptoms such as pain, swelling, bleeding at the 
site of the bite, ecchymosis and necrosis, which are common to most 
crotaline envenomings, the systemic bothropic syndrome observed 
in Central and South America is characterised by the development of 
consumption coagulopathies and spontaneous systemic bleeding, 
depending on venom components which affect clotting factors as 
well as haemorrhagins which damage vascular endotheliums 
(Barrantes et al, 1985; Kamiguti et al, 1991). On the other hand, 
apart from similar local signs, the severity of systemic envenoming 
by Bothrops lanceolatus in Martinique was correlated with the 
development of multiple cerebral infarctions and/or other major 
vessel occlusion that may appear within 8 hours to 7 days after the 
bite in approximately 30 to 40% of cases (Thomas etal, 1995, 1998). 
Infarctions can develop in patients who present initially with signs 
of moderate envenoming with normal blood clotting and low serum 
levels of venom antigens. The infarction process can involve several 
small vascular territories altogether, and is associated with the 
development of an isolated thrombocytopenia. Bogarin(V«/. ( 1999) 
demonstrated that Bothrops lanceolatus venom, obtained from 20 
specimens collected at different locations in Martinique, is devoid of 
thrombin-like enzymes and of in vitro coagulant and defibrinating 
activities, and is not coagulant when added to human citrated 
plasma, even at concentration as high as 100 ug/mL. These data 
suggest that thromboses observed in human B. lanceolatus enven- 
oming result from a toxin-linked vasculitis process rather than from 
a systemic procoagulant effect. However, the exact fhrombogenic 
mechanism responsible for these thromboses remains unexplained. 

The monophy ly of Bothrops lanceolatus and B. caribbaeus leads to 
the prediction that these snakes may share venom properties, which 
may in turn be of importance for the treatment of patients bitten by 
these snakes. In particular, do bites by B. caribbaeus result in a similar 
thrombotic syndrome as observed in B. lanceolatus? Bothrops 
caribbaeus envenoming was poorly documented until now. However, 
the case of a 32 year old man who was bitten in Saint Lucia and who 
subsequently developed multiple cerebral infarctions in the anterior 
and posterior cerebral artery territories was recently published (Nu- 
meric et al, 2002). The clinical presentation of this patient was 
identical to that of patients bitten by Bothrops lanceolatus. Thus, 
envenomings from these two species develop a unique systemic 
thrombotic syndrome, which differs fundamentally from the defibri- 
nation and bleeding syndrome that characterizes all other Bothrops 
asper-atrox complex envenomations. This example suggests that, at 
least in some cases, an understanding of the phylogeny of medically 
important snakes can help predict the syndrome of envenoming to be 
expected from a hitherto undocumented species. 



Our results also have implications for the conservation of the 
Antillean Bothrops. Our data show that both B. caribbaeus and B. 
lanceolatus represent relatively old, independent evolutionary line- 
ages, and not recent offshoots of widespread South American taxa. 
Conservation policy on their respective islands needs to take this into 
account. Although Lazell (1964) described both B. lanceolatus and 
especially B. caribbaeus as common (and Dowling, 1965, reported 
similar experiences for the latter), more recent workers have reported 
these snakes to be harder to find (Powell & Wittenberg, 1998). These 
observations indicate that B. caribbaeus and B. lanceolatus may have 
suffered a decline in population numbers over the last few decades, and 
that a reassessment of their conservation status should be a priority. 

Finally, this paper also represents an opportunity to clarify some 
confusion surrounding the nomenclature and synonymy of the 
Caribbean Bothrops. As noted by Hoge & Romano Hoge (1978/79) 
and subsequent authors, the St. Lucian lancehead was described 
under several different names by Gray (1842). Species of Bothrops 
described by Gray ( 1 842) include B. cinereus ('America'), B. sabinii 
('Demerara'). and B. subscutatus ('Demerara'). Gray (1849) also 
described B. affinis ('Demerara' and 'Berbice'). 

The types of B. sabinii and B. subscutatus were the specimens 
collected by Capt. (later Col.) Sabine discussed by Underwood 
(1993), and are unquestionably assignable to B. caribbaeus 
(Underwood, 1993; pers. obs.), of which the names B. subscutatus 
and B. sabinii therefore represent senior synonyms. However, the 
precedence of Garman's well-established name B. caribbaeus over 
Gray's disused names was formally established by Wuster (2000). 

The female type specimen of Bothrops cinereus, considered incertae 
sedis by Peters & Orejas-Miranda (1970) and conspecific with B. 
caribbaeus by Hoge & Romano Hoge (1978/79) and Powell & 
Wittenberg (1998), has 31 scale rows at midbody and 224 ventral 
scales. These counts are consistent with B. lanceolatus, but not with B. 
caribbaeus; B. cinereus is thus a junior synonym of B. lanceolatus. 
The sy ntypes of B. affinis are assignable to B. atrox. and are consistent 
with Guyanan populations of that species based on both scalation (24- 
27 dorsal scale rows, 1 89-200 ventrals) and colour pattern. 



Acknowledgements. We thank A. Malhotra, N.C. Giannasi and A. 
Tanasi (Office National des Forets de la Martinique) for help with sample 
acquisition, and C.J. McCarthy for access to the types of Gray's species of 
Bothrops. Finally, Garth Underwood provided enlightening information on 
Capt. Sabine's specimens, as well as being an inexhaustible font of knowledge 
on taxonomic matters of all kinds over many years. This study was funded by 
the Wellcome Trust (Research Career Development Fellowship to WW. and 
grant 057257//Z/99/Z), the EU (contracts TS3-CT9 1-0024 and IC1 8-CT96- 
0032), Fundacao Banco do Brasil, Fundacao de Amparo a Pesquisa do 
Estado de Sao Paulo (FAPESP) (grants 95/90 56-9, 97/2445-5 and 00/0 1 850- 
8), and the British Council (fellowship to MGS). 



REFERENCES 



Barrantes, A., Solis, V. & Bolaiios, R. 1985. Alteracion en los mecanismos de la 
coagulacion en el envenenamiento por Bothrops asper (terciopelo). Toxicon 23: 
399^07. 

Bogarin, G., Romero, M., Rojas, G., Lutsch, C., Casadamont, M., Lang, J., Otero, 
R. & Gutierrez, J. M. 1999. Neutralization, by a monospecific Bothrops lanceolatus 
antivenom, of toxic activities induced by homologous and heterologous Bothrops 
snake venoms. Toxicon 37: 551-557. 

Bremer, K. 1994. Branch support and tree stability. Cladistics 10: 295-304. 

Campbell, J.A. & Lamar, W.W. 1989. The Venomous Reptiles of Latin America. 
Comstock, Ithaca & London. 



106 



W. WUSTER ETAL. 



DeSalle, R. & Templeton, A.R. 1988. Founder effects and the rate of mitochondrial 

DNA evolution in Hawaiian Drosophila. Evolution 42: 1076-1084. 
Dowling, H.G. 1965. The puzzle of Bothrops; or, a tangle of serpents. Animal Kingdom 

68: 18-21. 
Gray, J.E. 1842. Synopsis of the species of rattle-snakes, or family of Crotalidae. 

Zoological Miscellany, Parts 2/3: 47-51. 
. 1849. Catalogue of Specimens of Snakes in the Collection of the British Museum. 

Edward Newman, London. 
Hedges, S. B. 1996. The origin of West Indian Amphibians and Reptiles, pp. 95-128. 

In: Powell, R. & Henderson, R.W. (eds) Contributions to West Indian Herpetology. 

A tribute to Albert Schwartz. SSAR, Ithaca. 
Henderson, R.W. & Hedges, S.B. 1995. Origin of West Indian populations of the 

geographically widespread boa Corallus enydris inferred from mitochondrial DNA 

sequences. Molecular Phylogenetics and Evolution 4: 88-92. 
Hillis, D.M. & Huelsenbeck, J.P. 1992. Signal, noise and reliability in phylogenetic 

analyses. Journal of Heredity 83: 189-195. 
, Mable, B. K. & Moritz, C. 1996. Applications of molecular systematics: the state 

of the field and a look to the future, pp. 5 15-543. In: Hillis, D.M., Moritz, C. & Mable, 

B. K. (eds) Molecular Systematics, 2 nd edition. Sinauer, Sunderland, Massachusetts. 
Hoge,A.R. 1952. Notaserpetologicas. Revalidacao de Bothrops lanceolata (Lacepede). 

Memorias do Instituto Butantan 24: 231-236. 
& Romano-Hoge, S.A.R.W.L. 1978/79. Poisonous snakes of the World. Part I. 

Checklist of the pit vipers, Viperoidea. Viperidae, Crotalinae. Memorias do Instituto 

Butantan 42/43: 179-310. 
Kamiguti A.S., Cardoso J.L.C., Theakston R.D.G., Sano-Martins I.S., Hutton 

R.A., Rugman F.R, Warrell D.A. & Hay, C.R.M. 1991. Coagulopathy and haem- 
orrhage in human victims of Bothrops jararaca envenoming in Brazil. Toxicon 29: 

961-972. 
Kumar, S. 1996. PHYLTEST: A Program for Testing Phylogenetic Hypothesis. 

Version 2. The Pennsylvania State University, University Park, Pennsylvania. 
Lazell, J.D. 1964. The Lesser Antillean representatives of Bothrops and Constrictor. 

Bulletin of the Museum of Comparative Zoology 132: 245-273. 
Malhotra, A. & Thorpe, R.S. 1999. Reptiles and Amphibians of the Eastern Carib- 
bean. MacMillan, London. 
& . 2000. The dynamics of natural selection and vicariance in the Dominican 

anole: comparison of patterns of within-island molecular and morphological diver- 
gence. Evolution 54: 245-258. 
Numeric, R, Moravie, V., Didier, M., Chatot-Henry, D., Cirille, S., Bucher, B. & 

Thomas, L. In press. Multiple cerebral infarctions following a snakebite by Bothrops 

caribbaeus. American Journal of Tropical Medicine and Hygiene. 
Peters, J.A. & Orejas-Miranda, B. 1970. Catalogue of the Neotropical Squamata. 

Part I. Snakes. Bulletin of the United States National Museum 297: 1-347. 
Pook, C. E., Wiister, W. & Thorpe, R. S. 2000. Historical biogeography of the western 

rattlesnake (Serpentes: Viperidae: Crotalus viridis), inferred from mitochondrial 

DNA sequence information. Molecular Phylogenetics and Evolution 15: 269-282. 
Posada, D. & Crandall, K.A. 1998. Modeltest: testing the model of DNA substitution. 

Bioinformatics 14: 817-818. 
Powell, R. & Wittenberg, R.D. 1998. Bothrops caribbaeus (Garman) Saint Lucia 

Lancehead. Catalogue of the American Amphibians and Reptiles (676): 1-4. 
Puorto, G., Salomao, M.G., Theakston, R.D.G., Thorpe, R.S., Warrell, D.A. & 

Wiister, W. 2001. Combining mitochondrial DNA sequences and morphological 

data to infer species boundaries: phylogeography of lanceheaded pitvipers in the 

Brazilian Atlantic forest, and the status of Bothrops pradoi (Squamata: Serpentes: 

Viperidae). Journal of Evolutionary Biology 14: 527-538. 
Salomao, M.G., Wiister, W., Thorpe, R.S., Touzet, J.-M. & B.B.B.S.P. 1997. DNA 

evolution of South American pitvipers of the genus Bothrops. pp. 89-98. In: Thorpe, 

R.S., Wiister, W. & Malhotra. A. (eds). Venomous Snakes: Ecology, Evolution and 

Snakebite. Clarendon Press, Oxford. 
, , & B.B.B.S.P. 1999. MtDNA phylogeny of Neotropical pitvipers of 

the genus Bothrops (Squamata: Serpentes: Viperidae). Kaupia 8: 127-134. 
Sandner Montilla, F. 198 1 . Una nueva subespecie de Bothrops lanceolatus (Lacepede, 

1789) Familia Crotalidae. Memorias Cientificas de Ofidiologi'a 6: 1-15. 
. 1990. Bothrops lanceolatus (Lacepede, 1789). Redescripcion amplia y bastante 

de la especie. Estudio completo sobre su taxonomia y caracteristicas morfologicas. 

Memorias Cientificas de Ofidiologia 10: 1 — 47. 
Shimodaira, H. & Hasegawa, M. 1999. Multiple comparisons of log-likelihoods with 

applications to phylogenetic inference. Molecular Biology and Evolution 16: 1114— 

1116. 
Swofford, D. L. 2001. PAUP* - Phylogenetic Analysis Using Parsimony (*and Other 

Methods). Beta version 4.0b8. Sinauer, Sunderland, Massachusetts. 
Templeton, A.R. 1983. Phylogenetic inference from restriction endonuclease cleavage 

sites maps with particular reference to the evolution of humans and the apes. 

Evolution 37: 221-244. 
Thomas, L., Tyburn, B., Bucher, B., Pecout, F., Ketterle, J., Rieux, D., Smadja, D., 

Gamier, D., Plumelle, Y. & Research Group on Snake Bite in Martinique. 1995. 

Prevention of thromboses in human patients with Bothrops lanceolatus envenoming 

in Martinique: failure of anticoagulants and efficacy of a monospecific antivenom. 

American Journal of Tropical Medicine and Hygiene 52: 419^426. 



, , Ketterle J., Biao T., Mehdaoui H., Moravie V., Rouvel C, Plumelle Y., 

Bucher B., Canonge D, Marie-Nelly C.A., Lang J. and other members of the 
Research Group on Snake Bites in Martinique. 1998. Prognostic significance of 
clinical grading of patients envenomed by snake Bothrops lanceolatus in Martinique. 
Transactions of the Royal Society of Tropical Medicine and Hygiene 92: 542-545. 

Thorpe, R. S., Malhotra, A., Stenson, A. & Reardon, J.T. in press. Adaptation and 
Speciation in Lesser Antillean Anoles. In Deickmann. U.. Metz. H.A.J.. Doebeli. M. 
& Tautz, D. (eds). Adaptive Speciation. Cambridge University Press. Cambridge. 

, McGregor, D.P., Cumming, A.M. & Jordan, W.C. 1994. DNA Evolution and 

colonization sequence of island lizards in relation to geological history: MtDNA 
RFLP, cytochrome b. cytochrome oxidase, 1 2srRNA sequence and nuclear RAPD 
analysis. Evolution 48: 230-240. 

Underwood, G. 1993. A new snake from St. Lucia. West Indies. Bulletin of the Natural 
History Museum (Zoology) 59: 1-9. 

Wiister, W. 2000. Precedence of names in wide use over disused synonyms or 
homonyms in accordance with Article 23.9 of the Code. Reptilia, Serpentes (1) 
Trigonocephalus caribbaeus Garman, 1887. Bulletin of Zoological Nomenclature 
57:9. 

, Salomao, M.G., Quijada-Mascarenas, J.A., Thorpe, R.S. & B.B.B.S.P. in 

press. Origin and evolution of the South American pitviper fauna: evidence from 
mitochondrial DNA sequence analysis. In: Schuett. G.W., Hoggren, M. & Greene. 
H.W. (eds). Biology of the Vipers. Biological Sciences Press, Carmel. Indiana. 

, , Duckett, G.J., Thorpe, R.S. & B.B.B.S.P. 1999. Mitochondrial DNA 

phylogeny of the Bothrops atro.x species complex (Squamata: Serpentes: Viperidae). 
Kaupia 8: 135-144. 

, , Thorpe, R.S., Puorto, G., Furtado, M.F.D., Hoge, S.A., Theakston, 

R.D.G. & Warrell, D.A. 1997. Systematics of the Bothrops atro.x species complex: 
insights from multivariate analysis and mitochondrial DNA sequence information, 
pp. 99-113. In: Thorpe, R.S.. Wiister. W. & Malhotra, A. (eds). Venomous Snakes: 
Ecology, Evolution and Snakebite. Clarendon Press, Oxford. 

Yang, Z., Goldman, N. & Friday, A. 1994. Comparison of models for nucleotide 
substitution used in maximum-likelihood phylogenetic estimation. Molecular Biol- 
ogy and Evolution 11: 316-324. 

Zhang, D.-X. & Hewitt, G.M. 1996. Nuclear integrations: challenges for mitochon- 
drial DNA markers. Trends in Ecology and Evolution 11: 247-251. 



Appendix 1 

Origin and vouchers of samples sequenced in this study. Institu- 
tional Codes for vouchers: IB = Instituto Butantan, Sao Paulo, 
Brazil, Herpetological Collection. FHGO = Fundacion Herpetologica 
Gustavo Orces, Quito, Ecuador. INHMT = Instituto de Higiene y 
Medicina Tropical 'L. Izquieta Perez', Guayaquil, Ecuador. ROM = 
Royal Ontario Museum, Toronto. WW = Wolfgang Wiister collec- 
tion. Collection numbers refer to preserved specimens unless 
otherwise stated. Photographs and/or morphological data for many 
unvouchered specimens are available from the first author. 

Bothrocophias microphthalmus: ECUADOR: ZamoraChinchipe: 
Cuencadel Rio Jamboe: Pumbami. FHGO 2566. Bothrops altematus: 
BRAZIL: Parana: Pinhao. IB 553 14. B. asper: Belize: Mile 38, West- 
ern Highway. WW 264. B. atrox: BRAZIL: Para: Santa Izabel. WW 
735. Maranhao: Sao Bento: WW 723. FRENCH GUYANA: Mana. 
WW 554. GUYANA: North West District: Baramita. ROM 22848. 
SURINAME: Coronie District: 7.5 km E. Totness. WW 537. B. 
brazili: ECUADOR: Morona Santiago: Macuma. FHGO 982. B. 
caribbaeus: SAINT LUCIA: Grande Anse. WW 144 WW 148. B. 
colombiensis: VENEZUELA: Guarico: Altagracia de Orituco. WW 
74. Falcon: San Francisco. J.L. Yrausquin, live coll. Guaibacoa. J.L. 
Yrausquin, live coll. B. insularis: BRAZIL: Sao Paulo: Ilha da Quei- 
mada Grande. Released aftersampling.B.y'araraca: BRAZIL: Parana: 
Piracuara. WW 926. B.jararacussu: BRAZIL: Sao Paulo: Cananeia. 
IB55313.fi. Zanceo/afK^:MARTINIQUE.Notvouchered.B. leucurus: 
BRAZIL: Bahia: Porto Seguro. IB 55480-1; Salvador. IB 55478. 
Espfrito Santo: Domingos Martins. IB 55557. B. marajoensis: BRA- 
ZIL: Para: Ilha de Marajo: 10 km NW Camara. WW 80. Santa Cruz do 
Arari: WW 943. B.pulcher. ECUADOR: ZamoraChinchipe: Estacion 
Cientifica San Francisco. FHGO live coll. 2 142. B. punctatus: ECUA- 
DOR: Pichincha: Pedro Vicente Maldonado. FHGO live coll. 2 1 66. B. 
taeniatus: ECUADOR: Morona Santiago: Macuma. FHGO 195. 



XX (ZS70&x,^ 



Bull. nut. Hist. Mus. Lond, (Zool.) 68(2): 107-11 



Issued 28 November 2002 



A contribution to the systematics of two 
commonly confused pitvipers from the Sunda 
Region: Trimeresurus hageni and T. 
sumatranus 



K. L. SANDERS, A. MALHOTRA AND R. S. THORPE 

School of Biological Sciences, University of Wales, Bangor, Gwynedd LL57 2UW, Wales, UK. 

SYNOPSIS. The systematics of two Southeast Asian green pitviper species, Trimeresurus hageni and T. sumatranus, are investi- 
gated by canonical variate analysis. Preliminary results reveal two morphological forms corresponding to mainly T. hageni in West 
Malaysia, Thailand and Singapore and T. sumatranus in Borneo. Allopatric populations of both taxa are examined from Sumatra. 
Geographic variation is present in both species, which are distinguished mainly by head scalation. but also by colour and pattern. 



INTRODUCTION 



Trimeresurus sumatranus (Raffles, 1822) and T. hageni (Lidth de 
Jeude, 1886) are closely related species, occupying low elevations 
in undisturbed forests and having largely overlapping ranges. The 
systematics of these species and their precise distribution is an area 
of long-standing confusion. Many workers assign both species to T. 
sumatranus by default (Tweedie, 1983; Lim, 1991; Jintakune, 1995; 
David and Vogel, 1996) and the status of T. hageni has been in 
dispute since its initial description (Lidth de Jeude, 1886; Lidth de 
Jeude, 1890;Boulenger, 1896; Brongersma, 1933). 

T. hageni was described as a separate species from T. sumatranus on 
the basis that only one or two supralabial scales are in contact with the 
subocular (compared with three in T. sumatranus), and the dark edges 
on head and body scales and dorsal cross-bands that are characteristic 
of T. sumatranus are not present (Lidth de Jeude, 1 886). The species' 
distribution is widely debated, but specimens from south Thailand, 
West Malaysia and Singapore are normally assigned to T. hageni, and 
specimens from Borneo are normally assigned to T. sumatranus 
(David and Vogel, 1996; Cox et al., 1998;StuebingandInger, 1999), 
but see Dring (1979) who placed specimens in the NHM collections 
from West Malaysia, southern Thailand and Sarawak in T. sumatranus. 
Both species are thought to occur on Sumatra and surrounding islands 
(Brongersma, 1933; Dring etai, 1989; Cox elai., 1998). 

There have been few attempts to resolve the systematics of T. 
hageni and T. sumatranus since their initial description; these have 
been based on small sample sizes and a traditional character-by- 
character approach (Boulenger, 1896; Brongersma, 1933). Given 
the levels of geographic, ontogenetic and sexual variation usually 
present in viper species (Wiister et al., 1992; Malhotra and Thorpe, 
1997), the systematics of these taxa is best approached using modern 
statistical methods based on a broad range of morphological 
characters. In this paper, we present preliminary results from an 
ongoing investigation of the systematics and interrelationships of T. 
hageni and T. sumatranus. 



MATERIALS AND METHODS 

We examined 78 specimens from museum collections in the United 
States, Europe and Malaysia (Figure 1). A total of 93 characters 



relating to scalation, colour and pattern were recorded for each 
specimen. Ventral scales were counted from head to vent, with the 
first ventral identified according to the method of Dowling ( 1951 ). 
The positions of scale reductions along the body (recorded as the 
number of the ventral or subcaudal scale opposite which it was 
situated) were transformed to percentage ventral scale (%VS) or 
caudal scale (%CS) position, in order to compensate for variation in 
ventral and subcaudal scale number. Male and female specimens 
were treated separately in all analyses to avoid bias caused by sexual 
dimorphism. 

Specimens were grouped by locality into operational taxonomic 
units (OTUs). Two groups dominated the analysis, one was com- 
prised of specimens from Thailand. West Malaysia and Singapore, 
and another was comprised of specimens from Borneo (Sabah and 
Sarawak). These groups were shown to be monophyletic by mole- 
cular analysis (unpublished data), which revealed a clear 
distinction between western specimens that lacked dorsal cross- 
bands and had at most two supralabials connected to the subocular 
scale, and eastern specimens that had dorsal cross-bands and had 
three supralabials in contact with the subocular scale. Molecular 
data was not available for specimens from Sumatra, and these 
were grouped individually to avoid combining sympatric species 
in one OTU. 

Each OTU was checked prior to further analysis using Princi- 
pal Component Analysis, which does not require that individuals 
be assigned groups prior to the analysis. The integrity of the 
OTUs was confirmed with the exception of one specimen from 
Betong (south Thailand), which had dark banding and in the PCA 
ordination was closest to the Borneo OTU. In subsequent analysis 
this specimen was grouped separately from the other western 
specimens. The OTUs used and their sample size for each sex is 
listed in Table 1. 

Variation between OTUs was tested for individual characters by 
means of one-way analysis of variance (ANOVA). Only characters 
showing significant between-OTU variation were used in subse- 
quent analyses. These are presented in Table 2. 

Canonical variate analysis (CVA) was used to investigate patterns 
of geographic variation between OTUs. This method maximises the 
separation between groups relative to variation within groups. It is a 
standard multivariate method and has been applied successfully to 
numerous models of geographic variation in reptiles (Wiister et al., 
1992; Thorpe et al., 1994; Daltry et ai, 1996). 



© The Natural History Museum. 2002 



108 



K.L. SANDERS ETAL. 




Fig. 1 Geographic origin of specimens used in multivariate analysis. S = Trimeresurus sumatranus; H = Trimeresurus hageni; U = unidentified 
specimens. Shading represents the known distribution of T. hageni and/or T. sumatranus. 



Table 1 List of OTUs and sample size for each sex. 



OTU 



Sample Size 
Males Females 



Thailand, West Malaysia, Singapore 

North Sumatra 1 (Medan) 

North Sumatra 2 (Medan) 

Central Sumatra 1 (Padang) 

Central Sumatra 2 (Padang) 

South Sumatra 1 (Palembang) 

Nias 

Siberut 

East Malaysia 

Betong (south Thailand) 

Total 



6 


15 


1 


1 





1 


1 


1 





1 





1 


1 


10 


3 


3 


4 


18 


1 





7 


51 



Museum Acronyms 

BMNH The Natural History Museum, London, formerly the 

British Museum (Natural History), London 
FMNH Field Museum of Natural History, Chicago 
IMR Institute of Medical Research, Kuala Lumpur 

KSP Sabah Park Zoological Museum, Mount Kinabalu 

National Park, Sabah 
MCZ Museum of Comparative Zoology, Harvard 

MHNG Museum d'Histoire Naturelle de Geneva, Switzerland 
NMBA Naturhistorisches Museum Basel, Switzerland 
NMW Naturhistorisches Museum Vienna, Austria 
QSMI Queen Saovabha Memorial Institute, Bangkok 
PH Perhelitan, Kuala Lumpur 

ZRC Raffles Museum of Biodiversity Research, National 

University of Singapore, Singapore 



West Malaysia and Singapore and those normally assigned to T. 
sumatranus from East Malaysia. The Siberut OTU and the single 
specimens from Nias and northern Sumatra are closest to the main- 
land T. hageni population. The specimens from Betong, Thailand 
and central Sumatra are closest to the Borneo OTU, but are well 
differentiated on CV2. 

Analysis of females also shows strong differentiation between the 
Thailand, West Malaysia and Singapore OTU and the Borneo OTU. 
The Siberut and Nias specimens are phenotypically close to T. 
hageni from Thailand, West Malaysia and Singapore. Specimens 
from north and south Sumatra are also closely affiliated to this 
mainland population. The specimens from central Sumatra are 
closest to the Borneo population along CV1, although are clearly 
differentiated on CV2. 

CVA analysis can be used to identify the characters that account 
for most variation between groups. In both sexes scalation characters 
were more important in distinguishing between the taxa than were 
characters relating to colour and pattern. The most important character 
is the fifth supralabial scale, which meets the subocular scale in T. 
sumatranus and in T. hageni is separated from the subocular by one 
scale. Also important is the frequent presence of an internasal scale 
in T. sumatranus, which is usually lacking in T. hageni. In addition, 
T. sumatranus has fewer supralabial scales and fewer scales between 
supraoculars than T. hageni. Our work verifies two of the original 
diagnostic characters used by Lidth de Jeude (1886) who described 
T. hageni as a distinct species that lacks dorsal cross-bands and has 
fewer supralabial scales in contact with the subocular scale. How- 
ever, we did not find dark edging on head and body scales to be a 
valid diagnostic character on the basis that T. hageni specimens from 
Nias have very strong dark edges on their head and body scales. 



RESULTS 



The CVA of males shows clear separation along the first canonical 
variate of specimens normally assigned to T. hageni from Thailand, 



DISCUSSION 



The results of this preliminary analysis reveal a major phenotypic 



TRIMERESURUS HAGENI AND T. SUMATRANUS 



109 



Table 2 Characters used for multivariate analysis of T. sumatranus & T. hageni. 



Characters 



Males 



Females 



1 . No. of ventral scales 

2. No. of subcaudal scales 

3. %VS position of reduction from 21 to 19 body scale rows 

4. %VS position of reduction from 19 to 17 body scale rows 

5. %DV position of reduction from 19 to 17 body scale rows 

6. %VS position of reduction from 1 7 to 15 body scale rows 

7. %CS position of reduction from 14 to 12 tail scale rows 

8. %DV position of reduction from 14 to 12 tail scale rows 

9. %CS position of reduction from 10 to 8 tail scale rows 

10. % DV position of reduction from 10 to 8 tail scale rows 

1 1 . %CS position of reduction from 8 to 6 tail scale rows 

12. %CS position of reduction from 6 to 4 tail scale rows 

13. No. of supralabial scales 

14. No. of sublabial scales 

15. No. of scales bordering the supraocular scales 

1 6. Minimum no. of scales separating the supraocular scales 

1 7. Maximum no. of scales separating the supraocular scales 

1 8. No. of internasal scales 

19. No. of scales separating the fourth supralabial scale form the subocular scale 

20. No. of scales separating the fifth supralabial scale form the subocular scale 

21 . No. of scales contacting the suboculars. excluding the preoculars and postoculars 

22. Average no. of scales between the first ventral scales and the anterior genial scales 

23. No. of scales between the last sublabial scales and first vental scales 

24. Presence of stripe on dorsal scale row one 

25. No. of scale rows involved in stripe 

26. Presence of postocular stripe 

27. No. of scale rows involved in postocular stripe 

28. Presence of dark edging on body scales 

29. No. of bands on body 

30. Mean no. of scales of three half bands on body 

3 1 . Mean no. of scales between three half bands on body 

32. Presence of dark edging on head scales 



* indicates significance value p=<0.05 (ANOVA) 



division in both sexes. This corresponds to T. sumatranus in Borneo, 
central Sumatra and southern Thailand and T. hageni in southern 
Thailand, West Malaysia, Singapore, north Sumatra, south Sumatra. 
Nias and Siberut. The species are best distinguished by head scalation. 
but can also be identified by colour and pattern. 

Geographic variation is also present at the intra-specific level. 
The Siberut and Nias specimens show stronger differentiation in 
males than in females. Their phenotypic similarity to mainland T. 
hageni is based mainly on scalation characters. Moreover, on the 
basis of colour and pattern, the Nias population is quite distinct with 
head and body scales strongly edged in black. Nias was last con- 
nected to Sumatra in the geologically recent past (c. 18.000 years 
ago), whereas Siberut has been isolated for around one million years 
(Dring et ah, 1989). The extent to which these populations have 
diverged from the mainland population will be investigated using 
molecular methods and may lead to taxonomic revisions. 

Sumatran populations are represented by few specimens, but 
these exhibit the same general pattern in males and females: T. 
sumatranus from central Sumatra appear to be strongly differenti- 
ated from the Borneo OTU, whereas T. hageni from north and south 
Sumatra are only weakly differentiated from the mainland OTU. 
This pattern will be tested when additional data becomes available. 
An analysis of the phylogenetic relationships of these populations, 
using mitochondrial sequence data, is also underway and should 
help to clarify their status. 



Acknowledgements. We thank our collaborators at the University of 
Science, Malaysia, and in particular Dr. Shahrul Anuar. We also thank the 
staff and curators of the following institutions for allowing us access to their 



specimens: BMNH. FMNH. IMR. KNP, MCZ, MHNG, NMBA, NMW. PH. 
QSM1.ZRC. This study was supported by the Natural Environment Research 
Council studentship to KLS (NER/S/A2O0O/03695), the Leverhulme Trust 
(F/I74/I and F/ 174/0). the Wellcome Trust (057257/Z/99/Z and 060384/Z/ 
00/Z), and the Darwin Initiative (162/6/65) with additional support for 
fieldwork from the Linnaean Society of London. Side. Bonhote, Omer- 
Cooper and Westwood Fund. 



REFERENCES 

Boulenger, G.A. 1 896. Catalogue of the Snakes of the British Museum (Natural 

History). Volume III. Containing the conclusion of the Colubridae.Amblycephalidae 

and Viperidae. London, British Museum (Natural History). 
Brongersma, L.D. 1933. Herpetological notes I -IX. Zoologische Mededeelingen 

Leiden 16: 1-29. 
Cox, M.J., van Dijk, P.P., Nabhitabhata, J. & Thirakhupt, K. 1 998. A photographic 

guide to snakes and other reptiles of Peninsular Malaysia, Singapore and Thailand. 

New Holland, UK. 
Daltry, J.C., Wiister, W. & Thorpe, R.S. 1996. Diet and snake venom evolution. 

Nature 379: 537-540. 
David, P. & Vogel, J. 1996. Snakes of Sumatra. Edition Chimaira. Frankfurt. 
Dowling, H. G. 1951. A proposed standard system of counting ventrals in snakes. 

British Journal of Herpetology 1: 97-99. 
Dring, J.C.M., McCarthy, C.J. & Whitten, A.J. 1 989. The terrestrial herpetofauna of 

the Mentawai islands. Indonsia. lndo-Malayan Zoology 6: 1 19-132. 
Dring, J.C.M.1979. Amphibians and Reptiles from northern Trengganu. Malaysia. 

with descriptions of two new geckos. Bulletin of the British Museum (Natural 

History), Zoology. 34(5): 181-241. 
Jintakune, P. & C. Lawan. 1995. Venomous Snakes of Thailand. The Thai Red Cross 

Society. Science Division. Bangkok. 
Lidth de Jeude, T. W. Van. 1886. Note X. On Cophias wagleri. Boie and Coluber 

sumatranus. Raffles. Notes Leyden Museum 8: 43-54. 



110 



K.L. SANDERS ETAL. 



13 

10 




9? 


A 
A A A 


5 






AA&^ 


n -j 












A^ 


-5 - 




♦ 




-10 






D n • 


-15 


<$> 




D + 


-20 








-25 


i 


i 


I I 



O Borneo 

A Thailand 

West Malaysia 
Singapore 

D Siberut 

+ Nias 

% north Sumatra 

<§> central Sumatra 

♦ Betong 



-80 



-60 



-40 



-20 



20 



IU 




■ 


5 


~ 











-5 




^A 


10- 


<§> 




15 


#> 




20 







-20 



O Borneo 

A Thailand 

West Malaysia 
Singapore 

D Siberut 

+ Nias 

9 north Sumatra 

<§> central Sumatra 

■ south Sumatra 



-15 -10 



10 



15 20 



Fig. 2 Canonical Variate Analysis of T. hageni and T. sumatranus populations (top = males; bottom = females). 






TRIMERESURUS HAGEN1 AND T. SUMATRANUS 



111 



1890. Note VIII. On a collection of snakes from Deli. Notes Leyden Museum 12: 

17-27. 
Lim, B. L. 1991. Poisonous Snakes of Peninsular Malaysia. The Malayan Nature 

Society, Malaysia. 
Malhotra, A. & Thorpe, R. S. 1997. New perspectives on the evolution of south-east 

Asian pitvipers (genus Trimeresurus) from molecular studies. Symposium of the 

Zoological Society of London (70): 1 15-128. 
Raffles. T. S. 1822. SirT. S. Raffles's Descriptive Catalogue of a Zoological Collection 

made in Sumatra. Transactions of the Linnean Society of London 3(2): 333-334. 
Stuebing, R.B. & Inger, R.F. 1999. A field guide to the snakes of Borneo. Natural 

History Publications, Borneo. 
Thorpe, R.S., Brown, R.P., Day, M., Malhotra, A., McGregor, D.R & Wuster, W. 

1994. Testing ecological and phylogenetic hypotheses in microevolutionary studies. 

Chapter 8. In: Eggleton, R & Vane-Wright, R. (eds). Phylogenetics and ecology. The 

Linnean Society of London. 
Tweedie, M. W. F. 1983. The Snakes of Malaya. Singapore National Printers Ltd., 

Singapore. 
Wiister, W., Thorpe, R.S. & Puorto, G. 1996. Systematics of the Bothrops atrox 

complex (Reptilia: Serpentes: Viperidae) in Brazil: A multivariate analysis. 

Herpetologica 52(2): 263-271. 
Wuster, W., Otsuka, S., Malhotra, A. & Thorpe, R.S. 1992. Population systematics 

of Russell's viper: A multivariate study. Biological Journal of the Linnean Society 

47:97-113. 



Appendix 1 

analysis 



Specimens used in morphological 



MUSEUM/FIELD REF 



LOCALITY 



QSMI 

QSMI 11190 
MHNG 2072.87 
MHNG 2072.89 
MHNG 1403.95 
MCZ 132799 
PH 
PH 
PH 

PH no.79 
PH no. 1 34 
FMNH 183787 
FMNH 183788 
FMNH 143948 
FMNH 138690 
FMNH 138689 
FMNH 148829 
FMNH 148830 
FMNH 138687 
FMNH 239948 
FMNH 239959 
FMNH 243943 
FMNH 230064 
FMNH 230063 



Pangna, Thailand 
Krabi, Thailand 
Surat Thani, Thailand 
Sural Thani, Thailand 
Singapore 

Kuala Lumpur, W. Malaysia 
Krau WR, W. Malaysia 
Tapa, W. Malaysia 
Selangor, W. Malaysia 
Kuala Jasin, W. Malaysia 
Ulu Gombak, W. Malaysia 
Janda Baik, W. Malaysia 
Ulu Gombak, W. Malaysia 
Selangor, W. Malaysia 
Kapit District, Sarawak 
Kapit District, Sarawak 
Kapit District, Sarawak 
Kapit District, Sarawak 
Kapit District, Sarawak 
Kota Mardu, Sabah 
Mendolong. Sabah 
Mendolong. Sabah 
Danum Valley. Sabah 
Danum Valley, Sabah 



SEX 



M 
M 
M 
M 
M 
F 
F 
F 
F 
F 
F 
F 
M 
M 
F 
F 
F 
F 
F 
F 
F 
F 
F 
F 



FMNH 239952 
FMNH 239950 
FMNH 239958 
FMNH 239957 
FMNH 239947 
FMNH 138688 
NMBA9179 
NMBA5108 
NMBA 22401 
NMW 28160.4 
NMW 28160.3 
NMW 28160.2 
NMW 28160.1 
NMW 28157.1 
NMW 28 156.1 
NMW 23909.2 
NMW 28159.4 
NMW 23909.3 
NMW 28159.2 
NMW 23909.4 
NMW 28159.1 
NMW 28158.1 
NMW 28 1 58.2 
BMNH 1936.9.12.3 
BMNH 1884.1.8.47 
BMNH 1884.1.8.46 
BMNH 1884.12.31.13 
BMNH 1884.12.31.14 
BMNH 1977.1237 
BMNH 1979.267 
BMNH 1979.268 
BMNH 1978.1879 
BMNH 1880.9.10.7 
BMNH 1936.9.12.5 
BMNH 1967.2290 
KSP 04361 
KSP 04362 
1MR 103649 
IMR 104270 
IMR 104271 
IMR 105684 
IMR 95995 
ZRC 2.2938 
ZRC 2.2937 
ZRC 2.2936 
KLS01129 
KLS0I119 
KLS 0008 
KLS0199 
AFS9815 
AFS 97b20 
AFS 0005 



Tenom, Sabah F 

Tenom, Sabah F 

Tenom, Sabah F 

Tenom. Sabah F 

Kota Mardu, Sabah M 

Kapit District, Sarawak M 

Nias, Indonesia M 

Palembang, Sumatra F 

Pangna. Thailand M 

Nias. Indonesia F 

Nias, Indonesia F 

Nias, Indonesia F 

Nias, Indonesia F 

Nias. Indonesia F 

Nias. Indonesia F 

Medan. Sumatra M 

Medan, Sumatra F 

Medan, Sumatra F 

Padang. Sumatra F 

Padang. Sumatra F 

Padang. Sumatra M 

Kedah, W. Malaysia F 

Kedah. W. Malaysia F 

Betong. Thailand M 

Nias, Indonesia F 

Nias. Indonesia F 

Nias. Indonesia F 

Nias. Indonesia F 

Siberut, Indonesia M 

Siberut, Indonesia F 

Siberut, Indonesia M 

G.Mulu, Sarawak F 

Singapore F 

Kuala Teku, W. Malaysia M 

G.Benom. W. Malaysia M 

Kota Mardu, Sabah F 

Bukit Tawau, Sabah F 

Bukit Lakai. W. Malaysia F 

Simpang, W. Malaysia F 

Ulu Gombak. W. Malaysia F 

Ulu Gombak. W. Malaysia F 

Simpang. W. Malaysia M 

Siberut. Indonesia M 

Siberut, Indonesia F 

Siberut, Indonesia F 

Poring, Sabah M 

Poring, Sabah M 

Endau Rompin NP, W. Malaysia F 

Perak, W. Malaysia M 

Songkhla, Thailand M 

Satun, Thailand F 

Kuala Lumpur. W. Malaysia F 



AFS/KLS indicate wild caught specimens examined under anaesthesia. 



XA {'±> '"*> ~\J 



Bull. not. Hist. Mux. Loud. (Zool.) 68(2): 113-121 



Issued 28 November 2002 



Underwood's classification of the geckos: a 21 st 
century appreciation 

ANTHONY P. RUSSELL 

Department of Biological Sciences, University of Calgary; 2500 University Drive N.W., Calgary, Alberta, 
Canada, T2N 1N4. email: anissell@ucalgary.ca 

AARON M. BAUER 

Biology Department, Villanova University, 800 Lancaster Avenue, Villanova, Pennsylvania 19085-1699, USA 

SYNOPSIS. The publication in 1 954 of Underwood's 'On the classification and evolution of geckos' was the first comprehensive 
attempt to understand the systematica, evolution and biogeography of this group of lizards. Combining the use of the exploration 
of novel characters with a global overview of geckos. Underwood erected hypotheses of relationship and patterns of distribution. 
In the 48 years since that landmark publication much has changed, but much has stayed the same. Underwood's division of geckos 
into four major clusters is still recognised today, although the sphaerodactyls are now regarded as a group derived from within 
the gekkonines, and the diplodactylines have been diminished by the removal of several genera and their placement in the 
gekkonines. The framework that Underwood established has resulted in generic and/or species level phylogenies being generated 
for the eublepharids, some sphaerodactyls, the carphodactyline diplodactylines and some clusters within the gekkonines. The 
latter group, because of its size, has remained intractable to detailed systematic analysis at the generic level, although the 
recognition of many discrete monophyletic clusters within the Gekkonidae (the Gekkoninae of Underwood) holds out the 
possibility that greater levels of intergeneric resolution are close to realisation. 

Underwood's initial approach to the systematic analysis of geckos was distinguished by its use of novel characters of the visual 
system that led to new insights. It is possible that the next breakthrough in higher level systematic analysis of geckos may again 
come from the exploitation of new character sources. Some examples of these possibilities are discussed. 



INTRODUCTION 



'[We] would like to make a distinction between how [Dr. Underwood! 
thought about the classification of [gekkotans] and what he thought, 
since in [our] view how a man thinks is far more important than what 
he thinks ... [we] suggest that how [Underwood] thought about 
classification survives untarnished to this day. Does what he thought 
about it bear critical scrutiny nearly fifty years later?' With these 
words (bracketed modifications aside) Garth Underwood (1971a) 
began his A Modern Appreciation of Camp's ''Classification of the 
Lizards.'' Nearly 50 years after its publication we here consider 
Underwood's classification of the Gekkota, its central position in the 
study of these lizards, its influence on subsequent work in the field 
and how its conclusions have been modified over the intervening 
decades. This work was Underwood's first substantial contribution 
to squamate evolution, preceding other major contributions to the 
systematics of pygopods (Underwood, 1957) and snakes 
(Underwood, 1967) and establishing the way he was to think about 
and employ character analysis in approaches to what had previously 
been regarded as rather intractable problems. 

Garth Underwood was interested in both the theory and practice 
of systematics and also in the evolutionary morphology of the 
organisms that he chose as the subjects of his systematic analyses. 
He was bold in embracing novel sources of data for his systematic 
analyses, put forward hypotheses of relationship in the hopes that 
they would be scrutinised and evaluated by others, and frequently 
returned to systematic problems that he had already published some 
years earlier to bring fresh insights and approaches. Specimens 
always figured prominently as a primary source of inspiration and 
new data. 



A comparison of the lizard families recognised by Boulenger 
( 1885), Camp (1923), and most modern workers (e.g., Macey e tal., 
1997; Harris et al, 1999) reveals almost no discrepancies. The 
gekkotan lizards, however, are an exception. Until the middle of the 
20"' century, lizard systematists variously recognised the 
Eublepharidae and Uroplatidae as entities distinct from the 
Gekkonidae. Different classificatory schemes reflected not only 
different interpretations of characters, but alternative views of the 
systematic meaning of novel morphologies. For most of the time 
after the description of the first gecko genera by Laurenti (1768), 
gecko systematics was dominated by alpha systematic treatments 
and the allocation of newly discovered species to an ever growing 
number of genera, defined chiefly by externally discernible digital 
features. This reliance on digital characters as almost the sole 
determinant of affinity resulted in the widespread recognition of 
composite genera constituted by digitally convergent taxa. Further, 
the focus on foot structure did little to resolve higher order relation- 
ships among gekkotans, as the digital characters then recognised 
suggested many discrete clusterings of species, but provided few 
putative links between them. 

Garth Underwood was led to the topic of gecko classification, 
which he ( 1954:469) characterised as 'far from stable', through his 
research on the reptilian eye. His earlier work on retinal morphology 
(Underwood 1951a) and pupil shape (Underwood 1951b) had both 
highlighted the distinctiveness of the gecko eye and suggested that 
ophthamological characters could be of use in the resolution of 
higher order relationships among the many gecko genera. 
Underwood's optimism that the eye could provide useful characters 
was bolstered by the then recent work of Bellairs (1948), who had 
conclusively demonstrated that the true eyelids of the eublepharid 
geckos were primitive to the derived condition of a well-developed 



© The Natural History Museum. 2002 



114 



A. P. RUSSELL AND A.M. BAUER 



brille and lack of moveable lids typical of other geckos. Further. 
Walls (1942) and Prince (1949) had examined the eyes of some 
geckos in their broader ophthamological treatments, suggesting 
avenues for further research. 

Walls' (1942) comprehensive treatment of the vertebrate eye led 
Underwood to hypothesise that this organ system could yield useful 
and stable associations of characters. The general recognition of the 
retinal characteristics of the eyes of geckos as evidence of secondary 
nocturnality, and a preliminary survey (Underwood 1951b) of the 
form of the pupil suggested that a more intensive survey of pupil 
form may provide a means by which gekkonids could be subdivided 
into more manageable and meaningful subsets reflective of their 
evolutionary history. Underwood (1954) set himself the task of 
surveying a moderately comprehensive collection of preserved 
geckos at the Museum of Comparative Zoology, Harvard Univer- 
sity, and to analyse the resulting data. He used these data to erect the 
first modern generic level analysis of gekkotan relationships. He 
recognised potential problems with character state interpretation 
caused by state of preservation and the limitations of a single- 
character classification, but nonetheless regarded pupil character 
states to be sufficiently discrete for the purpose of establishing a 
workable classification of geckos, which would be subject to modifi- 
cation as additional data became available. Werner (1977) later 
demonstrated that pupil shape and dilation change with differing 
light levels, and these observations have helped refine Underwood's 
(1954) initial conclusions (see below). 



UNDERWOOD'S CLASSIFICATION OF THE 
GECKOS 

Underwood (1954) recognised three families of gekkotan lizards. 
The Eublepharidae was characterised by true eyelids, the lack of a 
spectacle and vertical pupils reflective of the nocturnal adaptations 
of the family. The five genera he included were those subsequently 
placed by Kluge ( 1967a) in his Eublepharinae, and by Kluge (1987) 
and Grismer (1988) in the Eublepharidae. [Note: The current alloca- 
tion of taxa employed in this article is based upon Kluge, 2001]. 
Underwood (1954) considered Aeluroscalabotes as the most primi- 
tive member of the family. It has subsequently been regarded as the 
sister group of all remaining eublepharids (Grismer, 1988). 

Underwood's Sphaerodactylidae was supported by the presence 
of a round, diurnal-type pupil (or eliptical or straight vertical pupil in 
some cases), the existence of a fovea, and the presence of a specta- 
cle. He included five genera therein, corresponding to Kluge's 
(1967a) Sphaerodactylinae and later Sphaerodactylini (Kluge 1987, 
1995). 

All remaining genera were placed in the Gekkonidae, character- 
ised by a spectacle and lack of a fovea. Pupil shape was variable. 
Within the Gekkonidae he recognised two subfamilies, the 
Diplodactylinae and the Gekkoninae. The former had vertical pupils 
with straight margins, or circular pupils. He included 22 genera in 
this group. Among them are all of the genera now assigned to the 
Diplodactylinae by Kluge (1967a) except for Eurydactylodes, 
Pseudothecadactylus, and Crenadactylus. Underwood had doubts 
about the placement of the first of these genera (see below), speci- 
mens of which he had not examined himself, and changed its 
allocation the following year (Underwood 1955). Crenadactylus 
ocellatus was examined but was included with Phyllodactylus in the 
Gekkoninae by Underwood ( 1 954). The Diplodactylinae was subse- 
quently retained by Kluge (1987) and Bauer (1990a), although its 
affinities with the Pygopodidae were uncertain (see below). 



Stephenson and Stephenson (1956) regarded New Zealand geckos 
(Hoplodactylus and Naultinus) as the most primitive forms on the 
basis of Underwood's (1955) revised view that amphicoelous verte- 
bral centra are primitive within lizards and within the Gekkota. 
Furthermore, Stephenson (1960) rejected Underwood's (1954) 
ophthamalogical division of the Gekkonidae into two subfamilies as 
it was inconsistent with osteological characters, but neither 
Underwood nor Stephenson 'correctly' placed all Australian genera. 

Also included in Underwood's Diplodactylinae were several 
genera not now regarded as closely allied to the Australo-Pacific 
diplodactylines: Aristelliger, Chondrodactylus, Colopus, 
Gymnodactylus, Palmatogecko, Phelsuma, Ptenopus, Rhoptwpella, 
Rhoptropus, SaurodacTylus, and Teratoscincus. Four of these, 
Chondrodactylus, Colopus, Rhoptropus, and Palmatogecko, share 
many features in common with each other and with Pachydactylus 
(placed by Underwood [1954] in the Gekkoninae). Kluge (1967a) 
moved these taxa to the Gekkoninae. and Russell ( 1 972) and Haacke 
(1976) established the affinities of these forms as part of the Pachy- 
dactylus group (see below). 

Two other taxa, Rhoptwpella and Phelsuma, have also been 
regarded as being closely related to one another (see below). Both of 
these genera, as well as all remaining ones, were moved to the 
Gekkoninae by Kluge (1967a) and have remained there since, with 
Teratoscincus as the sister group of all other gekkonines. The 
affinities of Gymnodactylus have remained problematic (Abdala 
1988, 1996; Abdala and Moro 1996), as have those of Aristelliger 
(Russell and Bauer 1993), and Ptenopus (Bauer 1990b), whereas 
Saurodactylus has been considered allied to the sphaerodactyline 
lineage (Kluge 1995). Underwood's Phyllurus also included within 
it a species now assigned to the gekkonine genus Nactus. 

The Gekkoninae were characterised by Gekko-lype pupils or 
secondarily circular pupils. Underwood's (1954) Gekkoninae, al- 
though lacking the taxa mentioned above (and with the addition of 
Eurydactylodes, and Crenadactylus as Phyllodactylus ocellatus) 
otherwise included all of the genera placed in the group by Kluge 
(1967a). This grouping also included Uroplatus, which by virtue of 
a large suite of autapomorphic features had been accorded separate 
familial status by many previous workers (see Bauer and Russell 
1989 for a review). In this regard, Underwood's (1954) results were 
similar to those of Wellborn (1933), who had based her conclusions 
on osteological data. Underwood did not rely entirely on the pupil 
character, however, as Lygodactylus, with round pupils, was placed 
in the Gekkoninae on the basis of other (digital) similarities with 
Hemidactylus. 

Nine genera were not assigned to family or subfamily by 
Underwood. Five of these were unplaced due to lack of material. 
The remaining four were taxa with round pupils that were regarded 
as secondarily diurnal gekkonids, but which Underwood consid- 
ered, on the basis of existing data, could not be allocated to one or the 
other of his two subfamilies. Of the latter, one genus, Ancylodactylus, 
has been synonymized with another, Cnemaspis. The other two were 
Quedenfeldtia and Pristurus. Of the genera not examined. 
Ceramodactylus has since been subsumed in Stenodactylus, and 
Dravidogecko has been synonymized with Hemidactylus. 

Underwood also recognised some instances of convergence among 
geckos. Specifically he addressed the allocation of species of leaf- 
toed geckos (then chiefly distributed in Diplodactylus and 
Phyllodactylus), and bent-toed geckos (then mostly placed in 
Gymnodactylus). Among the leaf-toed geckos, pupil shape sug- 
gested the transfer of several species of African Diplodactylus to 
Phyllodactylus. These geckos are now regarded as members of the 
genus Urocotyledon (Kluge, 1 983) and are, as Underwood indicated, 
correctly assigned to the Gekkonidae rather than the Diplodactylidae. 






GECKO CLASSIFICATION 



15 



Diplodactylyus, as recognised by Underwood, corresponds to two 
currently recognised genera, Diplodactylus and Strophurus. His 
reconstituted Phyllodactylus included forms now placed in that 
genus as well as Asaccus, Afrogecko, Euleptes, Christinus, 
Crenadactylus, Paroedura and Urocotyledon (based on his list of 
specimens examined). He also separated Narudasia from 
Quedenfeldtia, and divided the then cosmopolitan Gymnodactylus 
into four genera: Gymnodactylus (restricted to South America), 
Phyllurus (corresponding to the current Phyllurus and Saltuarius, 
but also including the species vankampeni, now allocated to the 
gekkonine genus Nadus), Cyrtodactylus (including representatives 
of Cyrtodactylus, Geckoella, Tenuidactylus. Mediodactylus, Nactus), 
and Wallsaurus (the latter now synonymized with Homonota, a 
genus listed as unexamined by Underwood). 



STEPS TOWARDS FURTHER SYSTEMATIC 
RESOLUTION 

Underwood's ( 1 954) classification provided a springboard for sub- 
sequent systematic work on geckos. The four large units he 
established were 'corrected' by Kluge (1967a), but remained as the 
chief elements in Kluge's higher order treatment of the group. 
Moffat (1973) generally accepted Kluge's (1967a) allocation of 
genera to subfamilies but disagreed with his methodology and his 
pattern of subfamilial relationships. 

Eublepharidae 

The most stable unit has been the Eublepharidae. This group was 
retained intact by Kluge (1967a), although reduced to subfamilial 
rank. All subsequent researchers have accepted the monophyly of 
this group and more recent treatments have reflected the phylogenetic 
position of the Eublepharidae as the sister-group of all other gekkotans 
by again according it familial rank (e.g. Grismer 1988). Further, 
patterns of relationship within the eublepharids have been estab- 
lished at the generic and species levels (Grismer, 1988, 1991, 1994; 
Grismer et al. 1999; Olaet al.. 1999). In this instance. Underwood 
(1954) chiefly used primitive features in diagnosing the family (e.g. 
true eyelids present, etc.) but subsequent research has identified 
numerous synapomorphies that support the reality of this mono- 
phyletic unit (Grismer, 1988; Ota et ai, 1999). 

Sphaerodactylidae 

The Sphaerodactylidae of Underwood has remained unchanged in 
terms of generic content. Kluge (1967a) recognised the group as a 
subfamily and considered it to be highly derived, in contrast to 
Underwood ( 1 954), who interpreted it as a primitively diurnal group 
and a relatively early offshoot of the gekkotan lineage. Subsequently 
Kluge (1987) demonstrated that sphaerodactyls are derived from 
within gekkonines, confirming their monophyly while obviating 
their recognition as a higher order group, as such recognition would 
render the Gekkoninae paraphyletic. This arrangement also re- 
ceived support from reproductive characters including the restriction 
of the calcareous eggshell to gekkonines and sphaerodactylines 
(Bustard 1968; Werner 1972). Kluge (1995) later conducted an 
explicit investigation of the phylogeny of the sphaerodactyls, yield- 
ing a fully resolved generic level pattern for the group. Kluge ( 1 995) 
regarded the gekkonine Pristurus as the immediate sister group of 
the sphaerodactyls and considered Quedenfeldtia, Cnemaspis, 
Narudasia and Saurodactylus as other appropriate outgroup taxa for 
his analysis (see below). Of these outgroup genera, Underwood 



examined material of only Narudasia and Saurodactylus. Species 
level analyses within individual sphaerodactyl genera are ongoing 
and have been attempted for the largest genus, Sphaerodactylus 
(Hass 1991, 1996). 

Diplodactylinae 

The composition of the Diplodactylinae has changed most signifi- 
cantly. Kluge (1967a, b) removed a large number of genera from this 
group to the Gekkoninae, leaving only forms with parchment- 
shelled eggs in his Diplodactylinae, and provided a generic level 
hypothesis of relationships among the remaining forms. Bauer 
(1990a) erected a species level hypothesis of relationships among 
the Carphodactylini, one of two tribal groups established by Kluge 
(1967a). Additional hypotheses at the species level have been 
presented by Good et al. (1997) and Vences et al. (2001). The 
Diplodactylini, also established by Kluge (1967a), has yet to be 
investigated phylogenetically at the species level, although Kluge 
( 1967b) erected a generic level hypothesis of relationships and King 
( 1 987b) suggested a species level phylogeny of Diplodactylus based 
on several karyotypic characters. Underwood (1954) had purged the 
genus Diplodactylus of two taxa with Gekko-type pupils, rendering 
a cluster of taxa still accepted as monophyletic. However, he re- 
tained in Phyllodactylus the species ocellatus, which has since been 
recognised as a diplodacty line and placed in the genus Crenadactylus. 

Although the content of Underwood's (1954) Diplodactylinae as 
a whole has changed little, argument persists over patterns of 
internal relationship. In particular, the monophyly of the 
Carphodactylini has been called into question (Donnellan et al. 
1999) and the relationship of New Zealand taxa has also been re- 
evaluated (Chambers et al. 2001). King (1987b) and King and 
Mengden ( 1990), based on chromosomal data, argued that Oedura 
was more closely allied to the Carphodactylini than to other 
Diplodactylini. and that pygopods are also allied to the 
carphodactylines. Donnellan et al. ( 1999), based on molecular data 
( 12SRNA, c-mos), regarded the Diplodactylini , including Oedura. 
as monophyletic, but suggested that the Carphodactylini is 
paraphyletic. They found pygopods to be the sister group of all 
Diplodactylines. 

Patterns of relationship within the Diplodactylinae have further 
been complicated by the recognition that pygopods are more closely 
related to this group (or some component thereof) than to other 
gekkotans (Kluge 1987). On this basis, Kluge (1987) recognised a 
redefined Pygopodidae for the group that includes diplodactyline 
geckos plus pygopods. Good et al. (1997), based in part on argu- 
ments presented by Bauer (1990a), proposed an alternative higher 
level scheme, recognising the Diplodactylidae as a family level 
group. Based on the patterns of relationship retrieved by Donnellan 
et al. ( 1999), the Diplodactylidae and Pygopodidae are sister taxa. 

As mentioned above, the genus Eurydactylodes proved particu- 
larly problematic to Underwood ( 1 954) and he only included it in his 
Diplodactylinae in the following year (Underwood, 1955). For a 
variety of reasons, this genus has continued to be enigmatic, exhib- 
iting an odd mosaic of characteristics. Although Eurydactylodes 
appears to be a member of a monophyletic New Caledonian 
carphodactyline radiation (Bauer 1990a), it possesses a number of 
features that are problematic and, at least superficially, link it to 
other groups of geckos. One such feature is the tail-squirting appa- 
ratus. Members of this genus have caudal glands that secrete a sticky 
substance as a defensive mechanism. Such mechanisms have been 
widely reported in arthropods (Deslippe et al. 1996), and amphi- 
bians (Arnold 1982), but among amniotes have been noted only for 
geckos of the Australian diplodactyline genus Strophurus (Rosenberg 



116 



A.P. RUSSELL AND A.M. BAUER 



and Russell 1980) and Eurydactylodes (Bohme and Sering 1997). 
Although the secretion has not been characterised, it is likely similar 
to that of Strophurus spp., which is proteinaceous (Rosenberg etal. 
1984) and is effective in detering at least some small predators, such 
as spiders, which become entangled in the secretion (Minton 1982). 
However, both the anatomy of the gland and the ejection mechanism 
of secretion differ between the two gecko genera, suggesting that the 
apparatus in convergent (Bohme and Sering 1997). Eurydactylodes 
is also convergent with Strophurus in its bright yellow-orange 
mouth coloration. Most geckos have unpigmented buccal linings. 

Eurydactylodes also shares some features with gekkonid geckos. 
Most notable is the presence of extracranial endolymphatic sacs in 
the neck region, especially in juveniles and reproductive females. 
These calcium-storing structures frequently form conspicuous bulges 
on the necks of gekkonids, but in diplodactylids are intracranial and 
contain little calcium. Eurydactylodes is an exception in that very 
large sacs are often present, in some individuals artificially increas- 
ing the apparent size of the head (Bauer 1989). Perhaps related to 
this, the eggshells of Eurydactylodes, although similar in most 
regards to those of typical carphodactylines, are covered by a 
calcified outer surface (Bauer and Sadlier 2000), which otherwise 
typifies gekkonids (Bustard, 1968; Werner, 1972). 

Gekkoninae 

The Gekkoninae was the most heterogeneous and unwieldy of 
Underwood's higher order groups and it has remained largely 
intractable to this day. Indeed, as a result of the resolution of the 
content of the Diplodactyinae, the Gekkoninae has grown signifi- 
cantly. Further, the vast majority of all new or resurrected genera 
since 1954 are gekkonines. Underwood (1954) initiated the process 
of dismantling some of the larger gekkonid genera that he recog- 
nised as polyphyletic assemblages of digitally convergent taxa. In 
particular he addressed the composition of Phyllodactylus and 
Gymnodactylus, two of the largest and most cosmopolitan taxa. 

Subsequent reduction of Phyllodactylus occurred with the removal 
of Crenadactylus and its shift to the Diplodactylinae (Dixon and 
Kluge 1964), and the placement of several geographically coherent 
gekkonine leaf-toed forms into Paroedura (Dixon and Kroll 1974). 
Asaccus (Dixon and Anderson 1973), Urocotyledon (Kluge 1983), 
and Christinus (Wells and Wellington 1983). All remaining Old 
World leaf-toed geckos were removed from the now strictly American 
Phyllodactylus by Bauer et al. (1997), who erected Haemodracon, 
Dixonius, Afrogecko, Cryptactites and Goggia, and resurrected 
Euleptes. Nussbaum et al. (1998) further provided a new generic 
name for the elongate-bodied leaf-toed geckos of Madagascar, 
Matoatoa. Arnold and Gardner (1994) also provided a species level 
phylogeny for Asaccus, using a variety of Old and New World leaf- 
toed geckos as outgroup taxa, but without explicit justification. Both 
these authors and Nussbaum et al. (1998) suggested that at least 
some phyllodactyl taxa might be closely related. 

A similar dismantling of Gymnodactylus was begun by Underwood 
(1954), who removed Phyllurus to the Diplodactylinae and recog- 
nised the genera Gymnodactylus, Cyrtodactylus and Wallsaurus for 
a subset of the naked-toed geckos. Subsequently Golubev and 
Szczerbak (1981) and Szczerbak and Golubev (1984) divided the 
Old World forms placed by Underwood in Cyrtodactylus, which 
they regarded as polyphyletic, into several genera, including the 
Palearctic Tenuidactylus, Cyrtopodion, Mesodactylus, Carinato- 
gecko, Mediodactylus and Asiocolotes. Tropical forms were divided 
into Cyrtodactylus, Geckoella and Nactus (Kluge 1983). 

The effect of these actions has been to dismantle several larger, 
clearly polyphyletic groups and to instead recognise a larger number 



of smaller, but putatively monophyletic, genera. The problem re- 
mains, however, that relationships among these genera are poorly 
resolved. While the identification of monophyletic units is a neces- 
sary first step in the resolution of gekkotan relationships, the increase 
in the number of such units increases the sampling required in order 
to erect a hypothesis of relationship across all members of the group. 
This has been the major stumbling block in the phylogenetic inter- 
pretation of the Gekkoninae: any attempt to resolve relationships 
among some subset of genera of necessity requires an analysis of 
virtually all other genera. The sheer diversity of the group has been 
an impediment to its resolution. 

Despite the difficulty of determining relationships among 
gekkonines, some clusters of genera that appear to be monophyletic 
have been identified. These groups are chiefly those that share 
highly distinctive and generally restricted derived conditions. Thus, 
such groups have typically been identified on the basis of informa- 
tion intrinsic to themselves rather than on the basis of outgroup 
comparison. Indeed, when outgroup analysis has been attempted, 
the choice of outgroup has been based on geography (e.g., Joger 
1985; Bauer 1990b; Abdala 1996; Macey et al. 2000) or on some 
preconceived notion of similarity, usually based on digital anatomy 
(e.g., Arnold and Gardner 1994; Macey et al. 2000). Chromosomal 
characteristics of gekkonids are highly heterogeneous (King 1987c), 
but such variation may occur within genera and thus has contributed 
little to the resolution of higher order relationships. 

One of the most substantially supported subgroups of gekkonines 
is the Pachydacnlus group. This is a cluster of genera sharing the 
unique feature of hyperphalangy of digit I of both the manus and pes. 
The group includes the chiefly Mediterranean genera Tarentola and 
Geckonia and the southern Africa forms Pachydactylus, Rhoptropus, 
Chondrodactylus, Colopus, and Palmatogecko. Underwood (1954) 
recognised the relationship of all of these except Pachydactylus 
itself, placing them in the Diplodactylinae and identifying a peculiar 
pupil shape, the Rhoptropus-type, that all shared. Several species of 
Pachydactylus (e.g., P. austeni, P. kochi) are strikingly similar, even 
in external appearance, to Colopus and Palmatogecko. By chance, 
however. Underwood's (1954) list of taxa examined reveals that he 
did not examine any of these species. Hyperphalangy had previ- 
ously been identified in some members of the group by Wellborn 
(1933), but her sampling was inadequate to highlight the potential 
phylogenetic value of the feature. Russell (1972, 1976) and Haacke 
(1976) recognised the significance of hyperphalangy and argued 
convincingly that this was evidence of the relatedness of these taxa. 
Virtually all subsequent workers (Bauer 1990b, 2000; Kluge and 
Nussbaum 1995; Lamb and Bauer 2002; but see Joger 1985) have 
agreed that these seven genera (including collectively approxi- 
mately 80 species) form a monophyletic group. With closely related 
taxa thus identified, species level phylogenies have been possible 
within constituent genera (e.g., Rhoptropus: Bauer and Good 1996, 
Lamb and Bauer 2001; Pachydactylus: Lamb and Bauer 2000, 
2002). 

Other clusterings, although less well investigated, have also been 
proposed, although not necessarily tested. The Gekko group, con- 
sisting of Gekko, Gehyra, Hemiphyllodactylus, Eepidodactylus, 
Luperosaurus, Perochirus, Pseudogekko, and Ptychozoon, all share 
similarities of digital structure (Kluge 1968; Russell 1972, 1976) 
and are probably a monophyletic group, although particular patterns 
of intergeneric relationship remain untested. 

The large and heterogeneous genus Hemidactylus seems to be 
related to a number of much smaller genera that are also similar 
digitally, and are united by synapomorphies of size and shape of the 
intermediate phalanges (Russell, 1977a). Dravidogecko, for example, 
has been synonymized with Hemidactylus on the basis of digital 






GECKO CLASSIFICATION 



117 



morphology (Bauer and Russell 1995). In addition, Cosymbotus, 
Briba and Teraiolepis are also very similar and are almost certainly 
share a common ancestry with Hemidactylus, or are derived from 
within it. 

Bauer (1990b) found some evidence for the recognition of a 
Madagascan radiation including several genera of leaf and fan-toed 
geckos including Uroplatus, Ebenavia and Paroedura. Kluge and 
Nussbaum (1995) did not retrieve identical patterns of relationship, 
but these genera nonetheless grouped closely when only Afro- 
Malagasy geckos were included in the analysis. An expanded Indian 
Ocean lineage, including these taxa plus Ailuronyx, Blaesodactylus, 
Homopholis, and Geckolepis was retrieved by Bauer (1990b), al- 
though not by Kluge and Nussbaum (1995). 

Another putatively monophyletic group is the Lygodactylus com- 
plex (Pasteur 1964), which includes two additional genera, at least 
one of which, Millotisaurus, is probably derived from within 
Lygodactylus (Pasteur, 1995; Kriiger, 2001). Lygodactylus itself 
clustered with Phelsuma in analyses constrained to include only 
Afro-Malagasy genera (Bauer 1990b; Kluge and Nussbaum 1995). 
Kriiger (2001 ) also clustered Lygodactylus and Phelsuma together. 

Although some genera have been revised at the alpha level, and 
numerous new taxa have been erected, most revisions have merely 
proposed species groups, without providing explicit hypotheses of 
relationship (e.g. Pasteur 1964; Brown and Parker 1977; Nussbaum 
and Raxworthy 2000). These, like many of the other groups, share 
digital similarities and geographic cohesiveness. Among those gen- 
era for which some idea of relationships exist, there are several for 
which species level phylogenies have been proposed, including 
Uroplatus (Bauer and Russell 1989) and Gchyra (selected species 
only; King 1979, 1983). 

Rhoptropella has been associated with several different genera by 
different authors. Russell (1977b) used digital morphology to argue 
that it was in fact a Phelsuma, with no direct affinities to Rhoptropus, 
with which it had previously been associated (e.g. Boulenger 1 885). 
Russell and Bauer (1990) found additional support for this from 
histological investigations and Good and Bauer (1995) presented 
allozyme evidence for Rhoptropella' s links to Phelsuma. Both 
Bauer (1990b) and Kluge and Nussbaum (1995) found the two 
genera to be sister taxa when a generic analysis was conducted. 
Rosier (2001 ), discussing pholidosis, also concluded that Phelsuma 
and Rhoptropella are sister taxa. Roll (1999), however, using oph- 
thalmological and digital surface data, interpreted it as displaying 
features of both Rhoptropus and Phelsuma, which, if true, could 
suggest affinities between the chiefly African Pachydactylus group 
and the putatively monophyletic Indian Ocean complex. A variety 
of character types also suggest that Bogertia and Thecadactylus may 
be allied (Russell and Bauer 1988; Abdala and Moro 1996). 

Cnemaspis, Narudasia, Quedenfeldtia. Saurodactylus and 
Pristurus have been proposed as gekkonine taxa basal to the 
sphaerodactyl lineage (Arnold 1993; Kluge 1995), demonstrating 
the paraphyly of the Gekkoninae. Although Kluge (1995) did not 
claim any specific relationships among these taxa, his analysis did 
yield patterns in which Pristurus was the sister group of the 
sphaerodactyls, and Narudasia, Saurodactylus and Cnemaspis 
formed a clade. Arnold (1993) advocated the pattern ({((Pristurus, 
Quedenfeldtia) sphaerodactyls) Saurodactylus) Narudasia). Behav- 
ioural apomorphies unique to this cluster were documented by 
Rosier and Wranik (2001), who noted reproductive morphological 
apomorphies shared by Quedenfeldtia and the sphaerodactyls to the 
exclusion of Pristurus. Arnold (1993) provided a species level 
phylogeny for Pristurus. The African members of this group were 
also clustered together in an anlysis of Afro-Malagasy taxa by Kluge 
and Nussbaum (1995). Roll and Schwemer (1999) identified a 



unique crystallin ligand common to several of these taxa (plus 
Lygodactylus), that they interpreted as synapomorphic. This was 
subsequently found in Cnemaspis (Roll, in press), but whether this 
indicates affinity or convergence among secondarily diurnal forms 
remains to be determined. 

The naked-toed geckos have proved especially difficult to deal 
with. Szczerbak and Golubev (1984, 1986) provided evidence of 
relationship among some Palearctic forms, such as Tenuidactylus, 
Mediodacrylus, Asiocolotes, and Cyrtopodion. Macey et al. (2000) 
found evidence for the monphyly of Cyrtopodion and Mediodacrylus 
and hypothesized relationships among a small number of species in 
each group. The generic allocation of certain Himalayan members 
of the group has proved especially problematic (Khan 1993; Khan 
and Rosier 1999). 

Another group of naked-toed geckos including Agamura, Bunopus, 
Alsophyla.x, Crossobamon, Microgecko, and Tropiocolotes has been 
even less well investigated (Leviton and Anderson 1972; Szczerbak 
and Golubev 1977; Golubev 1984; Golubev and Szczerbak 1985). 
The New World naked toed forms, Gymnodactylus and Homonota, 
have been included in analyses by Abdala (1996) and Abdala and 
Moro ( 1 996) but these investigations included only South American 
gekkonines. Abdala ( 1988) also provided a species level phylogeny 
for Homonota (see also Vanzolini 1968). 

While some degree of resolution for the gekkonine taxa outlined 
above has been reached, certain other gekkonines remain enigmatic 
and without any sound indication of affinities. Teratoscincus is 
highly unusual in its morphology, and appears to be the sister group 
of all remaining gekkonines ( Kluge 1 987). A species level phylogeny 
for this group has been generated (Macey et al. 1 999). Stenodactylus 
has sometimes been considered to be allied to Teratoscincus (Kluge 
1 967a; Kluge and Nussbaum 1 995 ). but its position remains equivo- 
cal (Arnold 1980). 

Another perplexing padless genus is Ptenopus, a southern African 
endemic. Both Bauer (1990b) and Kluge and Nussbaum (1995) 
found little evidence for particular affinities, and constrained or 
retrieved a basal placement among African gekkonines. Ptenopus 
possesses a large number of autapomorphic traits (Haacke 1975; 
Rittenhouse et al. 1998; Russell et al. 2000). This mirrors the 
situation that plagued analyses of Uroplatus in that many features 
segregate these geckos from other taxa, but those traits that are 
shared are chiefly primitive ones. 

Four pad-bearing genera, which appear unrelated to one another 
and have no obvious affinities to previously discussed groups, are 
also problematic. These are Afroedura, Aristelliger, Calodactylodes, 
and Paragehyra. Paragehyra was long known from a single speci- 
men of a single species, but a second species was recently discovered 
(Nussbaum and Raxworthy 1994). The availability of additional 
material allowed the relationships of the genus to be investigated in 
more detail, but this has not yielded any definitive statements about 
its position within the Gekkoninae (Kluge and Nussbaum 1995), 
although Nussbaum and Raxworthy (1994) noted the similarity of 
the digits of this form to those of another enigmatic taxon, the West 
Indian Aristelliger. 

Russell (1972) grouped Afroedura and Calodactylodes in the 
same digitally defined cluster. Loveridge (1944) had initially segre- 
gated Afroedura from the Australian Oedura, and this was reflected 
in Underwood's (1954) placement of the genera in different sub- 
families. Some question as to the distinctiveness of these taxa 
remained, however, until Cogger (1964) conducted detailed osteo- 
logical comparisons. Despite some similarities in digital design, 
Russell and Bauer (1989) concluded that Calodactylodes and 
Afroedura were more likely convergent than related. Bauer and Das 
(2000) noted some superficial similarity and geographic proximity 



118 



A.P. RUSSELL AND A.M. BAUER 



to Asaccus, but again concluded that the relationships of 
Calodactylodes were obscure. 

Aristelliger was one of the taxa regarded as enigmatic by 
Underwood (1954). He placed it in the Diplodactylinae and re- 
garded it as an archaic form, possibly unable to compete with the 
gekkonines, which he regarded as more derived. Indeed, he regarded 
it as being a basal gekkonnid, retaining oil droplets in the eyes and 
displaying vertebral amphicoely. In part. Underwood's (1954) as- 
sessment of this genus may have been influenced by the fact that he 
was, at the time, based in the West Indies and had more information 
about it than most other geckos, and certainly more than any that he 
also placed in the Dioplodactylinae. Aristelliger has been employed 
in a variety of evolutionary (Hecht 1 952 ) and morphological (Ruibal 
and Ernst 1965) studies, probably because of ease of availability. 
These studies, however, have helped little to clarify the position of 
the taxon. Although it has rather complex external digital structure, 
anatomically it reveals a quite simple architecture. Thus more 
detailed studies of the digits (Russell 1976, 1979; Russell and Bauer 
1990. 1993) have not assisted in placing it with other genera that 
typically show a more complex anatomy. 



BIOGEOGRAPHIC AND EVOLUTIONARY 
IMPLICATIONS OF UNDERWOOD'S 
CLASSIFICATION OF THE GECKOS 

Underwood (1954) pioneered a comprehensive approach to gecko 
systematics. As a result of this, he was faced with issues of biogeog- 
raphy and evolution that begged an explanation. For geckos, this 
was essentially uncharted territory and the recognition of clusters, 
especially within his Gekkoninae, generated new biogeographic and 
evolutionary problems. Chief among these was the need to explain 
the biogeography and evolution of his Diplodactylinae. This proved 
especially challenging because, as noted above, this cluster of taxa 
later proved to be the least stable of Underwood's ( 1954) proposed 
units. 

Underwood (1954) interpreted eublepharids, with their scattered 
distribution, as an ancient radiation with its own specialisations, 
chiefly to arid conditions, rather than as a cluster of relicts. He 
viewed the eublepharids as the primary, ancient Northern Hemi- 
sphere radiation of the Gekkota. 

The sphaerodactylids were biogeographically non-problematic 
as all occur in the New World. Underwood ( 1 954) viewed them as an 
early New World offshoot of the Gekkota, based on his belief that 
they were primitively diurnal, retaining certain plesiomorphic 
lacertilian ophthalmological features. Kluge's (1967) demonstra- 
tion that the sphaerodactyls are derived from within the gekkonines, 
and subsequent recognition of secondary diurnality in the 
sphaerodactylines (Roll, in press) has resulted in a reinterpretation 
of sphaerodactyl biogeography and evolutionary history, with north 
African affinities being supported by more recent systematic inves- 
tigations (Arnold, 1993; Kluge, 1995). 

Underwood (1954) undertook to explain the distribution of the 
Diplodactylinae which, in his view, included a large core of Australo- 
Pacific taxa, but also genera from Africa and the Americas. He noted 
that no genus occupied more than one continent and that most genera 
had rather limited or patchy distributions. Only Aristelliger and the 
New Zealand taxa did not co-occur with Gekkonines. He felt that 
ovoviviparity might explain their ability to survive in New Zealand. 
In the case of Aristelliger, he noted that its occurrence was basically 
complementary to that of gekkonines, and suggested that it may 
have formerly had a broader distribution but had subsequently 



withdrawn in the face of competition with gekkonine geckos. He 
viewed the gekkonines as a more modern, expanding group that was 
displacing diplodactylines from areas of previous occupancy. He 
regarded New Caledonia as marking the periphery of the range of 
the gekkonines, with Lepidodactylus and Eurydactylodes being 
relatively recent invaders into diplodactyline (Rhacodactylus and 
Bavayia) territory. He believed that Phelsuma, being chiefly insular, 
diurnal, and arboreal, was ecologically segregated from the 
gekkonines with which it co-occurs. He regarded its occurrence in 
mainland East Africa as a recent event. Its arrival on islands of the 
Indian Ocean was hypothesised to be as a nocturnal stock, an 
offshoot of the southern African cluster of diplodactylines, with a 
subsequent change in life style enabling it to coexist with gekkonines. 
He regarded most continental diplodactylines as being terrestrial, 
with arboreal forms being peripheral. 

The foregoing rather tortuous scenario developed by Underwood 
(1954) to account for diplodactyline biogeography and evolution 
was the direct result of the recognition of, as it was formulated at the 
time, apolyphyletic assemblage. Removal of Aristelliger, Phelsuma 
and a variety of other taxa (see above) from the Diplodactylinae 
( Kluge, 1 967) and inclusion of Eurydactylodes within it (Underwood, 
1955) rendered biogeographic and evolutionary consideration of the 
remaining diplodactylids more tractable (Bauer. 1990a), but left the 
Gekkonidae (Underwood's Gekkoninae) yet more unwieldy. That 
some gekkonine genera were present on multiple continents sug- 
gested to Underwood (1954) that this was the dominant group. He 
recognised four major digital morphologies among gekkonines, and 
believed that each had reached most areas of the world and that most 
had radiated in situ in each area, giving rise to numerous regionally 
endemic genera. Thus, while expansion was an important theme in 
the evolution of gekkonines, there was significant within-region 
evolution as well. These ideas were obviously heavily influenced by 
those of Darlington ( 1 948 ) and by the idea of competitive exclusion 
(a more ecological than historical view). He noted the waif dispersal 
capabilities of some geckos and opined that this complicated the 
picture of dispersal via land bridges that served as his main para- 
digm. The issue of waif dispersal, though recognised as being 
restricted to certain taxa, remains to this day as a confounding factor 
in the interpretation of the evolution of gekkonid spatial patterns. 

Further systematic consideration (see above) has resulted in an 
increased complement of gekkonid genera, but has also resulted in 
some level of internal resolution, which, in turn, has influenced 
some aspects of biogeographic interpretation. For many regions, 
local radiations of monophyletic clusters of genera have been recog- 
nised, but resolution of pattern between these clusters remains 
poorly understood. 



CONCLUSIONS 

Underwood's (1954) systematic, biogeographic and evolutionary 
considerations of geckos marked the first attempt to comprehen- 
sively assess this circumglobal and highly diverse cluster. His 
analyses brought some degree of order to a previously very poorly 
understood set of problems, and his choice of ophthalmological 
characters as those of primary consideration resulted in the estab- 
lishment of a basic pattern that has survived to the present in 
modified form. Although Underwood (1968, 1970, 1971b, 1977a,b) 
revisited the gekkotan eye repeatedly, the promise of phylogenetic 
utility originally held out by ophthalmolgical data has not. until 
recently, been pursued. Roll (1995, 1997, 1999) and Roll and 
Schwemer (1999) have demonstrated that many diurnal geckos are 



GECKO CLASSIFICATION 



119 



unable to modify pupil shape and instead regulate light through 
absorbance by crystallins in the lens. Although Roll and Schwemer 
( 1999) assumed that the use of particular crystallins was likely to 
have evolved only once, there is no evidence that all diurnal gekkonids 
are allied (e.g. Phelsuma + Lygodactyhts and Sphaerodactylus + 
Quedenfeltia + Narudasia + Saurodactylus; Kluge and Nussbaum 
1995). This avenue of approach, however, suggests that at the 
anatomical and molecular level, data from the visual system may yet 
be of significance in assisting in the resolution of pattern between 
nocturnal and secondarily diurnal clusters of gekkonids (including 
sphaerodactyls). 

Despite attempts to move away from digital architecture as a 
primary means of identifying suprageneric clusters, this has contin- 
ued to play a role and has been instrumental, by way of examination 
of internal architecture, in assisting in the circumscription of a 
number of apparently monophyletic assemblages (Russell, 1976). 
Pedal anatomy remains a primary determinant of generic allocation 
and a major clue to potential higher order relationships (e.g., 
Nussbaum and Raxworthy 1994). 

Changes in generic alignment and more modern views of plate 
tectonics have necessitated a rethinking of Underwood's (1954) 
biogeographic hypotheses. Essentially the eublepharids appear to 
represent an ancient Laurasian radiation, in keeping with Under- 
wood's (1954) ideas. The remaining gekkotans are now regarded as 
being of Gondwanan origin and to consist of an essentially east 
Gondwanan diplodactylid radiation and a west Gondwanan gekkonid 
radiation, with the latter having given rise, in turn, to the New World 
sphaerodactyls. 

Interpretation of patterns of relationship must now deal with the 
recognition that the age of the Gekkota is much greater than was 
believed in 1954 and that many genera might be quite ancient. 
Hence, generic body plans may have been established for very long 
periods, making them rather discrete from one another and render- 
ing it difficult to erect hypotheses of relationship. Even among the 
sphaerodacyls. generic differentiation is estimated to have occurred 
as much as 40 million years ago (Hass 1 99 1 ). King ( 1 987a, 1 987b), 
on the basis of chromosomal and immunological data correlated 
with tectonic history of the Australian region, estimated a minimum 
divergence of 66 my between the two major clades of diplodactylines, 
and at least 120 my for the origin of the gekkotans. 

Despite the magnitude of the problem, only patterns of relationship 
within the rather amorphous Gekkonidae ( Underwood's Gekkoninae) 
remain relatively unassailed. Even here, however, large, circumglobal 
unwieldy genera have been broken into smaller, more geographically 
circumscribed taxa and there is now an opportunity to begin to make 
inroads into the determination of the patterns of interrelationship of 
suprageneric clusters of gekkonid taxa. This may best be broached by 
taking exemplars, appropriately selected (Bininda-Emonds et al. 
1998) from the putative clusters and the enigmatic genera, and 
investigating a combination of morphological and molecular data. 
Given the magnitude of the problem, this will be an iterative process 
and will necessitate frequent cross-checking within and between 
clusters. The boldness of Garth Underwood's approach will have to 
be adopted in selecting novel sources of data to allow new approaches 
to be taken and insights to be revealed. 



REFERENCES 

Abdala, V. 1 988. Analisis cladi'stico de las especies del genero Homonota (Gekkonidae). 

Revista Espanola de Herpetologia 12: 55-62. 
1996. Osteologia craneal y relaciones de los geconinos sudamericanos 

(Reptilia:Gekkonidae). Revista Espanola de Herpetologia 10: 41-54. 



& Moro, S. 1996. Cranial musculature of South American Gekkonidae. Journal 

of Morphology 229: 59-70. 
Arnold, E.N. 1 980. A review of the lizard genus Stenodactylus (Reptilia: Gekkonidae). 

Fauna of Saudi Arabia 2: 368-404. 

1993. Historical changes in the ecology and behaviour of semaphore geckos 

(Pristurus, Gekkonidae) and their relatives. Journal of Zoology. London 229: 353- 
384. 

& Gardner, A.S. 1994. A review of the Middle Eastern leaf-toed geckoes 

(Gekkonidae: Asaccus) with descriptions of two new species from Oman. Fauna of 
Saudi Arabia 14: 424-441. 

Arnold, S.J. 1982. A quantitative approach to antipredator performance: salamander 
defense against snake attack. Copeia 1982: 247-253. 

Bauer, A.M. 1989. Extracranial endolymphatic sacs in Eurydactylodes (Reptilia: 
Gekkonidae). with comments on endolymphatic function in lizards in general. 
Journal of Herpetology 23: 172-175. 

1990a. Phylogenetic systematics and biogeography of the Carphodactylini (Rep- 
tilia: Gekkonidae). Bonner zoologische Monographien 30: 1-217. 

1 990b. Phylogeny and biogeography of the geckos of southern Africa and the 

islands of the western Indian Ocean: a preliminary analysis, pp.275-284. In: Peters. 
G. and Hutterer, R. (eds). Vertebrates in the Tropics. Zoologisches Forschungsinstitut 
und Museum A. Koenig, Bonn. 

2000. Evolutionary scenarios in the Pachydacrylus group geckos of southern 

Africa: new hypotheses. African Journal of Herpetology 48: 53-62. 

& Das, I. 2000. A review of the gekkonid genus Calodacrylodes (Reptilia: 

Squamata) from India and Sri Lanka. Journal of South Asian Natural History 5: 25- 
35. 

& Good, D.A. 1 996. Phylogenetic systematics of the day geckos, genus Rhoptropus 

(Reptilia: Gekkonidae). of south-western Africa. Journal of Zoology. London 238: 

635-663. 
& Russell, A. P. 1989. A systematic review of the genus Uroplatus (Reptilia: 

Gekkonidae). with comments on its biology. Journal of Natural History 23: 169- 

203. 

& 1995. The systematic relationships of Dravidogecko anamallensis 

(Giinther 1875). Asiatic Herpetological Research 6: 30-35. 

& Sadlier, R.A. 2000. The Herpetofaunu of New Caledonia. Society for the Study 

of Amphibians and Reptiles, Ithaca. NY. 

, Good, D.A. & Branch, W.R. 1997. A taxonomy of the Southern African leaf- 
toed geckos (Squamata: Gekkonidae), with a review of Old World 'Phyllodacrylus' 
and the description of five new genera. Proceedings of the California Academy of 
Sciences 49: 447^197. 

Bellairs, A.d'A. 1948. The eyelids and spectacle in geckos. Proceedings of the 
Zoological Society of London 118: 420-425. 

Bininda-Emonds, O.R.P., Bryant, H.N. & Russell, A.R 1998. Supraspecific taxa as 
terminals in cladistic analysis: implicit assumptions of monophyly and errors in 
phylogenetic inference. Biological Journal of the Linnean Society 64: 101-133. 

Bonnie. W. & Sering, M. 1997. Tail squirting in Eurydactylodes: independent evolu- 
tion of caudal defensive glands in a diplodactyline gecko (Reptilia. Gekkonidae). 
Zoologischer Anzeiger 235: 225-229. 

Boulenger, G.A. 1885. Catalogue of the Lizards in the British Museum (Natural 
History). 2nd ed. Vol. 1. Trustees of the British Museum, London. 

Brown, W.C. & Parker, F. 1 977. Lizards of the genus Lepidodacty/us (Gekkonidae) from 
the Indo- Australian Archipelago and the islands of the Pacific, with descriptions of new 
species. Proceedings of the California Academy of Sciences. 4'" ser. 61: 253-265. 

Bustard, H.R. 1968. The egg-shell of gekkonoid lizards: a taxonomic adjunct. Copeia 
1968: 162-164. 

Camp. C.L. 1923. Classification of the lizards. Bulletin of the American Museum of 
Natural History 48: 89-481. 

Chambers, G.K., Boon, W.M., Buckley, T.R., & Hitchmough, R.A. 2001. Using 
molecular methods to understand the Gondwanan affinities of the New Zealand 
biota: three case studies. Australian Journal of Botany 49: 377-387. 

Cogger, H.G. 1964. The comparative osteology and systematic status of the gekkonid 
genera Afroedura Loveridge and Oedura Gray. Proceedings of the Linnean Society 
of New South Wales 89: 364-372. 

Darlington, P.J., Jr. 1948. The geographical distribution of cold-blooded vertebrates. 
Quarterly Review of Biology 23: 1-26, 105-123. 

Deslippe, R.J., Jelinski, L., & Eisner, T. 1 996. Defense by use of a proteinaceous glue: 
woodlice vs. ants. Zoology 99 (1995/96): 205-2 10. 

Dixon, J.R. & Anderson, S.C. 1973. A new genus and species of gecko (Sauria: 
Gekkonidae) from Iran and Iraq. Bulletin of the Southern California Academy of 
Sciences 72: 155-160. 

& Kluge, A.G. 1964. A new gekkonid lizard genus from Australia. Copeia 1964: 

174-180. 

& Kroll, J.C. 1974. Resurrection of the generic name Paroedura for the 

phyllodactyline geckos of Madagascar, and description of a new species. Copeia 
1974: 24-30 

Donnellan, S.C, Hutchinson, M.N. & Saint, K.M. 1 999. Molecular evidence for the 
phylogeny of Australian gekkonoid lizards. Biological Journal of the Linnean 
Society 67: 97-118. 



120 



A.P. RUSSELL AND A.M. BAUER 



Golubev, M.L. 1984. Structure of the genus Tropiocolotes (Reptilia. Gekkonidae) [in 
Russian]. Vestnik Zoologii 1984(6): 12. 

& Szczerbak, N.N. 1981. Carinatogecko gen. n. (Reptilia, Gekkonidae) a new 

genus of gecko lizards from Southwest Asia [in Russian]. Vestnik Zoologii 1981(5): 
34^U. 

& Szczerbak, N.N. 1985. On the relationships of two Palearctic gecko genera: 

Tenuidactylus I Agamura (Reptila. Gekkonidae) [in Russian], pp. 59-60. In: Prob- 
lems ofHerpetology. Sixth All-Union Herpetological Conference Abstracts. Tashkent. 
Nauka, Leningrad. 

Good, D.A. & Bauer, A.M. 1995. The Namaqua day gecko revisited: allozyme 
evidence for the affinities of Phelsuma ocellata. Journal of the Herpetological 
Association of Africa 44: 1-9. 

, & Sadlier, R.A. 1997. Allozyme evidence for the phylogeny of giant New 

Caledonian geckos (Squamata: Diplodactylidae: Rhacodactylus), with comments on 
the status of R. leachianus henkelii. Australian Journal of Zoology 45: 317-330. 

Grismer, L.L. 1988. Phylogeny, taxonomy, classification, and biogeography of 
eublepharid geckos, pp. 369^-69. In: Estes, R. and Pregill. G. (eds). Phylogenetic 
Relationships of the Lizard Families: Essays Commemorating Charles L. Camp. 
Stanford University Press, California. 

1991. Cladistic relationships of the lizard Eublepharis turcmenicus (Squamata: 

Eublepharidae). Journal of Herpetology 25: 251-253. 

1994. Phylogeny, classification, and biogeography of Goniurosaurus kuroiwae 

(Squamata: Eublepharidae) from the Ryukyu Archipelago, Japan, with description of 
a new subspecies. Zoological Science 11: 319-335. 

, Viets, B.E. & Boyle, L.J. 1999. Two new continental species of Goniurosaurus 

(Squamata: Eublepharidae) with a phylogeny and evolutionary classification of the 

genus. Journal of Herpetology 33: 382-393. 
Haacke, W.D. 1975. The burrowing geckos of Southern Africa, 1. (Reptilia: 

Gekkonidae). Annals of the Transvaal Museum 29: 197-243. 
Haacke, W.D. 1976. The burrowing geckos of Southern Africa, 5. (Reptilia: 

Gekkonidae). Annals of the Transvaal Museum 30: 71-89. 
Harris, D.J., Sinclair, E.A., Mercader, N.L., Marshall, J.C. & Crandall, K.A. 1999. 

Squamate relationships based on c-mos nuclear DNA sequences. Herpetological 

Journal*. 147-151. 
Hass, C.A. 1991 . Evolution and biogeography of West Indian Sphaerodactylus (Sauria: 

Gekkonidae): a molecular approach. Journal of Zoology, London 225: 525-561. 
1996. Relationships among West Indian geckos of the genus Sphaerodactylus. a 

preliminary analysis of mitochondrial 16s ribosomal RNA sequences, pp. 175-194. 

In: Powell, R. & Henderson. R.W. (eds). Contributions to West Indian Herpetology: 

A Tribute to Albert Schwartz. Society for the Study of Amphibians and Reptiles, 

Ithaca, New York. 
Hecht, M.K. 1952. Natural selection in the lizard genus Aristelliger. Evolution 6: 112- 

124. 
Joger, U. 1985. The African gekkonine radiation — preliminary phylogenetic results. 

based on quantitative immunological comparisons of serum albumins, pp.479^1-94. 

In: Schuchmann, K.-L. (ed). African Vertebrates: Systematics, Phylogeny and 

Evolutionaiy Ecology. Zoologisches Forschungsinstitut und Museum A. Koenig. 

Bonn. 
Khan, M.S. 1993. A new angular-toed gecko from Pakistan, with remarks on the 

taxonomy and a key to the species belonging to the genus Cyrtodactylus (Reptilia: 

Sauria: Gekkonidae). Pakistan Journal of Zoology 25: 67-73. 
& Rosier, H. 1999. Redescription and generic redesignation of the Ladakhian 

gecko Gymnodactylus stoliczkai Steindachner. 1969 [sic]. Asiatic Herpetological 

Research 8: 60-68. 
King, M. 1979. Karyotypic evolution in Gehyra (Gekkonidae: Reptilia). I. The Gehyra 

variegata-punctata complex. Australian Journal of Zoology 27: 373-393. 

1983. Karyotypic evolution in Gehyra (Gekkonidae: Reptilia). II. The Gehyra 

australis complex. Australian Journal of Zoology 31: 723-741. 

1987a. Origin of the Gekkonidae: chromosomal and albumin evolution suggests 

Gondwanaland. Search 18: 252-254. 
1987b. Chromosomal evolution in the Diplodactylinae (Gekkonidae: Reptilia). I. 

Evolutionary relationships and patterns of change. Australian Journal of Zoology 

35: 507-531. 
1987c. Monophyleticism and polyphyleticism in the Gekkonidae: a chromosomal 

perspective. Australian Journal of Zoology 35: 641-654. 
& Mengden, G. 1990. Chromosal evolution in the Diplodactyinae (Gekkonidae: 

Reptilia). II. Chromosomal variability between New Caledonian species. Australian 

Journal of Zoology 38: 219-226. 
Kluge, A.G. 1967a. Higher taxonomic categories of gekkonid lizards and their 

evolution. Bulletin of the American Museum of Natural Histoty 135: 1-60. pis. 1-5. 

1967b. Systematics, phylogeny, and zoogeography of the lizard genus 

Diplodactylus Gray (Gekkonidae). Australian Journal of Zoology 15: 1007-1 108. 

1968. Phylogenetic relationships of the gekkonid lizard genera Lepidodactylus 

Fitzinger, Hemiphyllodactylus Bleeker, and Pseudogekko Taylor. The Philippine 
Journal of Science 95:331-352. 

1983. Cladistic relationships among gekkonid lizards. Copeia 1983: 465^75. 

1987. Cladistic relationships in the Gekkonoidea (Squamata, Sauria). Miscellane- 



ous Publications. Museum of Zoology. University of Michigan (173): i-iv + 1-54. 
1 995. Cladistic relationships of sphaerodactyl lizards. American Museum Novitales 

(3139): 1-23. 

2001. Gekkotan lizard taxonomy. Hamadryad 26:1-209. 

& Nussbaum, R.A. 1995. A review of African-Madagascan gekkonid lizard 

phylogeny and biogeography (Squamata). Miscellaneous Publications, Museum of 

Zoology, University of Michigan (183): i-iv + 1-20. 
Kriiger, J. 2001. Die madagassischen Gekkoniden. Tiel II: Die Geckos der Gattung 

Lygodactylus Gray, 1864 (Reptilia: Sauria: Gekkonidae). Gekkota 3: 3-28. 
Lamb, T. & Bauer, A.M. 2000. Relationships of the Pachydactylus rugosus group of 

geckos (Reptilia; Squamata: Gekkonidae). African Zoology 35: 55-67. 
& 2001 . Mitochondrial phylogeny of Namib day geckos (Rhoptropus) based 

on cytochrome b and 16S rRNA sequences. Copeia 2001: 775-780. 
& 2002. Phylogenetic relationships of the large-bodied members of the 

African lizard genus Pachydactylus (Reptilia: Gekkonidae). Copeia. in press. 
Laurenti, J.N. 1768. Specimen Medicum, Exhibens Synopsin [sic] Reptilium 

Emendatam cum Experimentis circa Venena et Antidota Reptilium Auslriacorum. 

Trattnern, Vienna. 
Leviton, A.E. & Anderson, S.C. 1972. Description of a new species of Tropiocolotes 

(Reptilia: Gekkonidae) with a revised key to the genus. Occasional Papers of the 

California Academy of Sciences 96: 1-7. 
Loveridge, A. 1 944. New geckos of the genera Afroedura. new genus, and Pachydactylus 

from Angola. American Museum Novitales 1254: 1— I. 
Macey, J.R., Ananjeva, N.B., Wang, Y. & Papenfuss, T.J. 2000. Phylogenetic 

relationships among Asian gekkonid lizards formerly of the genus Cyrtodactylus 

based on cladisitic analyses of allozymic data: monophyly of Cyrtopodion and 

Mediodactylus. Journal of Herpetology 34: 258-265. 
, Larson, A., Ananjeva, N.B., & Papenfuss, T.J. 1997. Replication slippage may 

cause parallel evolution in the secondary structures of mitochondrial transfer RNAs. 

Molecular Biology and Evolution 14: 30-39. 
, Wang, Y., Ananjeva, N.B., Larson, A. & Papenfuss, T.J. 1999. Vicariant 

patterns of fragmentation among gekkonid lizards of the genus Teratoscincus 

produced by the Indian collision: a molecular phylogenetic perspective and an 

area cladogram for Central Asia. Molecular Phylogenetics and Evolution 12: 

320-332. 
Mutton, S.A. 1982. Highlights in herpetology: it's not all snakes. Proceedings of the 

Indiana Academy of Sciences 92: 70-76. 
Moffat, L.A. 1973. The concept of primitiveness and its bearing on the phylogenetic 

classification of the Gekkota. Proceedings of the Linnean Society of New South 

Wales 97: 275-301. 
Nussbaum, R.A. & Raxworthy, C.J. 1994. The genus Paragehyra (Reptilia: Sauria: 

Gekkonidae) in southern Madagascar. Journal of Zoology. London 232: 37-59. 
& 2000. Systematic revision of the genus Paroedura Giinther (Reptilia: 

Squamata: Gekkonidae). with the description of five new species. Miscellaneous 

Publications. Museum of Zoology, University of Michigan (189): i-iv + 1-26. 
, & Pronk, O. 1998. The ghost geckos of Madagascar: a further revision of 

the Malagasy leaf-toed geckos (Reptilia, Squamata, Gekkonidae). Miscellaneous 

Publications, Museum of Zoology, University of Michigan (186): i-iv + 1-26. 
Ota, H., Honda, M., Kobayashi, M., Sengoku, S. & Hikida, T. 1999. Phylogenetic 

relationships of eublepharid geckos (Reptilia: Squamata): a molecular approach. 

Zoological Science 16: 659-666. 
Pasteur, G. 1964. Recherchessurl'evolutiondeslygodactyles. lezards Afro-Malgaches 

actuels. Travaux de ITnstitut Scientifique Cherifien. Serie Zoologie 29:1-132 + 12 

plates. 
1995. Biodiversite et reptiles: diagnoses de sept nouvelles especes fossiles et 

actuelles du genre de lezards Lygodactylus (Sauria. Gekkonidae). Dumerilia 2: 1-21. 
Prince, J.H. 1949. Visual development 1. E. & S. Livingstone, Edinburgh. 
Rittenhouse, D.R., Russell, A.P. & Bauer, A.M. 1998. The larynx and trachea of the 

barking gecko, Ptenopus garrulus maculatus (Reptilia: Gekkonidae) and their 

relation to vocalization. South African Journal of Zoology 33: 23-30. 
Roll, B. 1995. Crystallins in lenses of gekkonid lizards (Reptilia. Gekkonidae). 

Herpetological Journal 5: 298-304. 
1 997. Photoreceptors of diurnal and nocturnal geckos. In: Eisner, N. and Wassle, 

H. (eds), Proc. 25 Gottingen Neurobiology Conference Vol. II, 502. Thieme, Stutt- 
gart. 
1999. Biochemical and morphological aspects of the relationship of the Namaqua 

day gecko to Phelsuma and Rhoptropus (Reptilia, Gekkonidae). Zoology 102: 50- 

60. 
In press. Cnemaspis: tertiarily diumal (Reptilia. Gekkonidae). Herpetological 

Journal. 
& Schwemer, J. 1999. L-Crystallin and vitamin A, isomers in lenses of diurnal 

geckos. Journal of Comparative Physiology A 185: 51-58. 
Rosenberg, H.I. & Russell, A.P. 1980. Structural and functional aspects of tail 

squirting: a unique defense mechanism of Diplodactylus (Reptilia: Gekkonidae). 

Canadian Journal of Zoology 58: 865-881. 
, & Kapoor, M. 1984. Preliminary characterization of the defensive secre- 
tion of Diplodactylus (Reptilia: Gekkonidae). Copeia 1984: 1025-1028. 






GECKO CLASSIFICATION 



121 



Rosier, H. 2001. Zur Rumpfbeschuppung der Gattung Phelsuma Gray, 1825: 
Zoogeographische Aspekte (Sauna: Gekkonidae). Gekkota 3: 47-73. 

& Wranik, W. 200 1 . Bemerkungen zur Fortpflanzungsbiologie von Geckonen — 

2. Untersuchungen zur Reprpoduktion von Pristurus obsli Rosier & Wranik. 1999 
und Pristurus sokotrunus Parker. 1938 im Vergleich zu anderen Geckos (Sauria: 
Gekkonidae). Gekkota 3: 125-182. 

Ruibal, R. & Ernst, V. 1965. The structure of the digital setae of lizards. Journal of 
Morphology 117: 271-294. 

Russell, A.P. 1 972. The foot ofgekkonid lizards: a study in comparative and functional 
anatomy. Unpublished Ph.D. thesis. University of London. England. 

1 976. Some comments concerning interrelationships amongst gekkonine geckos. 

pp.217-244. In: Bellairs. A.d'A. & Cox, C.B. (eds). Morphology and Biology of 
Reptiles. Linnean Society Symposium Series 3. Academic Press. London. 

1977a. The phalangeal formula of Hemidactylus Oken, 1817 (Reptilia: 

Gekkonidae): a correction and a functional explanation. Anatomia Histologic! 
Embryologia 6: 332-338. 

1977b. The genera Rkoptropus and Phelsuma (Reptilia: Gekkonidae) in Southern 

Africa: a case of convergence and a reconsideration of the biogeography of Phelsuma. 
Zoologica Africana 12: 393-408. 

1979. Parallelism and integrated design in the foot structure of gekkonine and 

diplodactyline geckos. Copeia 1979: 1-21. 

& Bauer, A.M. 1988. Paraphalangeal elements ofgekkonid lizards: a compara- 
tive survey. Journal of Morphology 197: 221-240. 

& 1989. The morphology of the digits of the golden gecko. Calodactylodes 

aureus (Reptilia: Gekkonidae) and its implications for the occupation of rupicolous 
habitats. Amphibia-Reptilia 10: 125-140. 

& 1990. Digit I in pad-bearing gekkonine geckos: alternate designs and the 

potential constraints of phalangeal number. Memoirs of the Queensland Museum 
29:453^172. 

& 1993. Aristelliger Cope. Caribbean geckos. Catalogue of American 

Amphibians and Reptiles 565. 1^4. 

, Rittenhouse, D.R. & Bauer, A.M. 2000. Laryngotracheal morphology of Afro- 

Madagascan geckos - a comparative survey. Journal of Morphology 245: 241-268. 
Stephenson, N.G. I960. The comparative osteology of Australian geckos and its 

bearing on their morphological status. Transactions of the Royal Society of New 

Zealand 84: 341-358. 
& Stephenson, E.M. 1956. The osteology of the New Zealand geckos and its 

bearing on their morphological status. Transactions of the Royal Society of New 

Zealand 84: 341-358. 
Szczerbak, N.N. & Golubev, M.L. 1977. Systematica of the Palearctic geckos (genera 

Gymnodactylus, Bunopus, Alsophylax) [in Russian]. Trudy Zoologiiskogo Institute! 

Akudemiya Nauk USSR 74: 120-133. 
& 1984. On generic assignement of the Palearctic Cyrtodactylui lizard 

species (Reptilia, Gekkonidae) [ in Russian]. Vestnik Zoologii 1984(2): 50-56. 



& 1986. Gecko Fauna of the USSR and Contiguous Regions [in Russian]. 

Naukova Dumka, Kiev. 
Underwood, G. 1951a. Reptilian retinas. Nature 167: 183. 

195 lb. Pupil shape in certain geckos. Copeia 1951: 211-212. 

1954. On the classification and evolution of geckos. Proceedings of the Zoologi- 
cal Society of London 124: 469^192. 
1955. Classification of geckos. Nature 175: 1089. 

1957. On lizards of the family Pygopodidae. A contribution to the morphology 

and phylogeny of the Squamata. Journal of Morphology 100:207-268. 

1967. A Contribution to the Classification of Snakes. British Museum (Natural 

History) Publication No. 653. Trustees of the British Museum (Natural History), 
London. 179pp. 

1968. Some suggestions concerning vertebrate visual cells. Vision Research 8: 

483-488. 

1970. The eye. pp. 1-97. In: Gans. C. & Parsons, T. (eds). Biology of the Reptilia, 

vol. X. Academic Press, London. 
1971a. A modern appreciation of Camp's 'Classification of the lizards.' Pp. vii- 

xvii In Camp, C.L., Classification of the Lizards. Society for the Study of Amphibians 

and Reptiles, Lawrence. Kansas. 

1971b. Vertebrate comparative anatomy at the cellular level: visual cells and 

phylogeny. pp. 64-69. In: Hecht. M. (ed). Vertebrate Evolution: Mechanism and 
Process. Report of the NATO Advanced Study Institute, Robert College, Istanbul. 
Turkey 4-15 August, 1969. NATO. 

1977a. Simplification and degeneration in the course of evolution of squamate 

reptiles. Colloques internationaux C.N.R.S. N° 266 - Mecanismes de la 
Rudimentation des Organes chez les Embryons de Vertebres. pp.34 1-351. 

1977b. Comments on phyletic analysis of gekkotan lizards. NATO Advanced 

Studies Institute Series (Life Sciences) 14: 53-55. 

Vanzolini, P.E. l968.Geography of the South American Gekkonidae. Arquivos de 

Zoologia 17: 85-112. 
Vences, M„ Henkel, F.-W. & Seipp, R. 2001. Molekulare Untersuchungen zur 

Phylogenie und Taxonomie der Neukalcdonischen Geckos der Gattung Rhacodactylus 

(Reptilia: Gekkonidae). Sulamandra 37: 73-82. 
Walls, G.L. 1942. The Vertebrate Eye. Cranhrook Institute Scientific Bulletin 19. 

Bloomfield Hills, Michigan. 
Wellborn, V. 1933. Vergleichende ostcologische Untersuchungen an Geckoniden, 

Eublepharidenand Uroplatiden. Sitzungsberichte der Gesellschaft Naturforschender 

Freunde zu Berlin 1933: 126-199. 
Wells, R.W. & Wellington, C.R. 1983. A synopsis of the class Reptilia in Australia. 

Australian Journal <>/ Herpetolog) 1: 73-129. 
Werner, Y.L. 1972. Observations on eggs ol cublepharid lizards, with comments on the 

evolution of the Gekkonoidea. Zoologische Mededelingen 47: 21 1-224. pi. I. 
1977. Ecological comments on some gekkonid lizards of the Namib Desert, South 

West Africa. Madoqua 10: 157-169. 



XX (2S7\IS.2>) 



Bull. nat. Hist. Mus. Lond. (Zool.) 68(2): 123-130 



Issued 28 November 2002 



The skull of the Uropeltinae (Reptilia, 
Serpentes), with special reference to the otico- 
occipital region 

OLIVIER RIEPPEL 

Department of Geology, The Field Museum. 1 400 S Lake Shore Drive, Chicago, IL 60605-2496, U.S.A. e-mail: 
rieppel@fieldmuse.um. org 

HUSSAM ZAHER 

Universidade de Sao Paulo, Institute de Biociencias, Departamento de Zoologia, Rua do Matao, Travessa 14, 
Cidade Universitaria, 05508-900, Sao Paulo, SP, Brazil 

SYNOPSIS. The skull anatomy ot'uropeltines is reviewed, and new data is presented on the highly derived otico-occipital region. 
A phylogenetic analysis of uropeltine interrelationships using parsimony is performed using characters derived from skull 
structure. The basal position of the genus Melanophidium is confirmed; Pseudotyphlops is a relatively derived uropeltine. in spite 
of its relatively large size. The monophyly of the genera Melanophidium and Rhinophis requires further testing. 



INTRODUCTION 



MATERIAL EXAMINED 



Uropeltinae (Nopcsa 1923. The name is here used as by Kluge 1991, 
Fig. 4; see also Cundall et al. 1993) remain an enigmatic group of 
basal alethinophidian snakes. This is largely due to their burrowing 
habits and restricted distribution, and the consequent scarcity of 
material available in public repositories. We have studied the 
uropeltine skulls from the collections of The Natural History Mu- 
seum, London, which permitted us to review the highly derived skull 
structure in this monophyletic clade of snakes. 

The first detailed description of a uropeltine skull was given by 
Baumeister (1908) in a monograph on the genus Rhinophis. Peculi- 
arities of the cranio-vertebral joint in the group were dealt with by 
Williams ( 1 959), Underwood ( 1 967), and Hoff stetter & Gasc ( 1 969). 
Some aspects of the skull of uropeltines were described by Rieppel 
(1977. 1978, 1983), Bellairs & Kamal (1981), and Wever (1978), 
but none of these studies addressed details of the morphology of the 
otico-occipital complex. The lower jaw of uropeltines was described 
by Rieppel & Zaher (2000). In their detailed analysis of the cranial 
anatomy and phylogenetic relationships of Anomochilus, Cundall & 
Rossman (1993), and Cundall et al., (1993), comment on various 
aspects of the skull structure of uropeltines, and their functional as 
well as phylogenetic significance. In particular. Cundall & Rossman 
(1993; see also Cundall & Greene 2000) recognized a fundamen- 
tally different design of skull adaptation to burrowing habits in 
scolecophidians and uropeltines (Fig. 1 ). The phylogenetic relation- 
ships of uropeltines within snakes were discussed in cladistic terms 
by Cundall et al. (1993), Scanlon & Lee (2000), andTchernov etal, 
(2000). Only one study has appeared so far that dealt with uropeltine 
interrelationships, based on microcomplement fixation techniques 
(Cadle et al. 1990). In this paper, we review the skull anatomy of 
uropeltines, adding new detail to previous descriptions (Rieppel 
1977, 1978) and providing new data on the detailed morphology of 
the otico-occipital complex. These morphological characters are 
used in a phylogenetic analysis of uropeltine interrelationships, 
which will be compared to the results obtained by Cadle et al. 
(1990). This study is presented in honor of Dr. Garth Underwood, 
who more than 20 years ago introduced the senior author to the study 
of the skull of 'henophidian' snakes. 



The present study is based on the investigation of the skull of the 
following taxa (generic names used in the manuscript refer only to 
the specimens here listed), arranged as outgroup taxa and in-group 
(uropeltine) taxa. 

Institutional abbreviations 

BMNH, British Museum (Natural History), now The Natural His- 
tory Museum; FMNH. Field Museum of Natural History, now The 
Field Museum. 

Outgroup taxa: Anilius scytale (FMNH 35683); Cylindrophis 
maculatus (BMNH 1930.5.8.48); Cylindrophis ruffits (FMNH 
179033); Boa constrictor (FMNH 22435, 22438); Calabaria 
reinhardtii (FMNH 31372); Candoia aspera (FMNH 13915); 
Candoia h. australis (FMNH 22997); Lichanura roseofusca (FMNH 
31565); Python molurus (FMNH 223198); Tropidophis pardalis 
(FMNH 233). 

In-group taxa: Melanophidium punch/turn (BMNH 1930.5.8.119); 
Melanophidium wynaudense (BMNH 1930.5.8.124-125); Platy- 
plecturus madurensis (BMNH 1930.5.8.111); Plecturus perroteti 
(BMNH 1930.5.8.105); Pseudotyphlops philippinus (BMNH 
1978.1092); Rhinophis drummondhayi (BMNH 1930.5.8.67-68); 
Rhinophis sanguineus (BMNH 1930.5.8.59); Teretrurus rhodo- 
gaster (BMNH 1930.5.8.98); Uropeltis woodmansoni (BMNH 
1930.5.8.73-74); 

Abbreviations used in the figures 

ang, angular; bo, basioccipital; com, compound bone; d, dentary; 
ec, ectopterygoid; f, frontal; Is, laterosphenoid; m, maxilla; n, nasal; 
oc, otic capsule; op-eo, opisthotic-exoccipital; p, parietal; pi, pala- 
tine; pm, premaxilla; po.vc, posterior opening of Vidian canal; prf, 
prefrontal; pro, prootic; pro.c, prootic canal; pro.r, preorbital ridge; 
pt, pterygoid; q, quadrate; sm, septomaxilla; so, supraoccipital; sp, 
splenial; tr.f.r, transverse frontal ridge; v, vomer; II, optic foramen; 
V,, trigeminal foramen (maxillary branch); V 3 , trigeminal foramen 
(mandibular branch); VII, facialis foramen; X, jugular foramen; 
XII, hypoglossal foramen. 



© The Natural History Museum, 2002 



124 



O. RIEPPEL AND H. ZAHER 




Fig. 1 The skull and mandible of Pseudotyphlops philippinus (BMNH 
1978.1092) in left lateral view. 



GENERAL ASPECTS OF THE SKULL 

The premaxilla of uropeltines is characterized by a single premaxil- 
lary foramen (Fig. 2). The vomerine processes of the premaxilla 
meet the vomer in a well-defined contact. The premaxilla of 
uropeltines shows characteristic variation within the group (Rieppel 
1977; Cundall & Rossman 1993). The anterior margin of its trans- 
verse process is more or less evenly rounded in Melanophidiwn, 
which correlates with a gentle anteromedial curvature of the anterior 
end of the maxilla (Fig. 2A). The two bones closely approach each 
other, or barely establish contact. This genus therefore retains a 
plesiomorphic configuration of the snout, which is also inferred to 
be present in Platyplecturus (the specimen BMNH 1930.5.8.111 
lacks the premaxilla, but retains the maxilla which shows an 
anteromedially curved anterior end), and which represents a con- 
dition similar to that seen in Anomochilus (Cundall & Rossman 
1993). The other uropeltines have a similar premaxilla, which 
carries an anteriorly projecting, bipartite rostrum. The straight 'trans- 
verse' processes point posterolaterally, and meet the straight maxilla 



in a shizarthrosis (Cundall & Rossman 1993; Fig. 2B-E). These two 
elements define the lateral margins of the strongly 'telescoped' 
(Haas 1930), i.e., tapering and pointed snout (see also Cundall & 
Rossman 1993, Fig. 25B). 

The maxilla of basal alethinophidians carries an anterior medial 
process (Rieppel 1977; Scanlon & Lee 2000), which is particularly 
well developed in uropeltines, where it participates in the formation 
of a broadly overlapping contact between maxilla, premaxilla, and 
vomer. In Melanophidiwn, the anterior medial process of the maxilla 
is not engaged in any such contact, but freely underlaps the 
septomaxilla. In Pseudotyphlops, the anterior medial process of the 
maxilla overlaps a medially extending horizontal flange of the 
transverse process of the premaxilla in a complex, interlocking 
premaxillary - maxillary contact (Fig. 2B, D-E). Rhinophis and 
Uropeltis are unique in that the anterior medial process of the 
maxilla forms a well-defined sutural contact with an anterior lateral 
process of the vomer in front of the opening for Jacobson's organ. 

The medial or choanal processes of the palatines of uropeltines 
are broad, arching over the choanal tubes and projecting ventrally 
again medial to the choanal tubes. Their ventral tips are embraced 
anteriorly by the posterior ends of the vomers, as is also the case in 
other basal alethinophidians (Cundall & Rossman 1993; Cundall et 
al. 1993; Rieppel 1983; Fig. 2, 3). The parasphenoid forms a sagittal 
interchoanal process that lies between the choanal processes of the 
palatines, as is also the case in Anomochilus and Cylindrophis 
(Cundall & Rossman 1993; Cundall et al. 1993). The dorsal lamina 
of the nasal is variously developed in uropeltines, but tends to be 
relatively broad and notched anterolaterally in species with a rounded 
snout, but slender and tapering to a fine tip in species with a strongly 
telescoped snout. 

The snout complex is suspended from the rest of the skull at the 
naso-frontal joint (see Rieppel 1978. for details) and through the link 
provided by the prefrontal (Fig. 3). The maxilla of uropeltines 
carries a well-developed ascending process that is in a firm planar 
(i.e., not interdigitating) contact with the prefrontal. Medial to the 
ascending process, the superior alveolar nerve canal is open dorsally 
in uropeltines, appearing as a groove on the dorsal surface of the 




Fig. 2 A-E The palate in uropeltine snakes. A, Melanophidiwn wynaudense (BMNH 1930.5.8.124); B, Pseudotyphlops philippinus (BMNH 1978.1092) 
C, Rhinophis sanguineus (BMNH 1930.5.8.59); D-E, Pseudotyphlops philippinus (BMNH 1978.1092). 



SKULL OF UROPELTINAE 



125 



mnr**inni»yP>y.:F^^&mmi V ti)U, i." 'A 'l*L 




Fig. 3 The snout complex of Pseudotyphlops philippinus ( BMNH 1978. 1092). 



maxilla, a unique condition among snakes (Fig. 4). The suspension 
of the snout complex from the braincase is more elaborate in 
uropeltines than it is in other basal alethinophidians (Rieppel 1978). 

The parietal of uropeltines forms distinct anterolateral, i.e., 
supraorbital processes which may or may not contact the prefrontal 
(Fig. 3). In Cylindrophis maculatus and in Anomochilus (Cundall & 
Rossman 1993), as well as in Melanophidium punctatum, the 
supraorbital process of the parietal participates in the suspension of 
the prefrontal. In other uropeltines, the contact between parietal and 
prefrontal may be reduced or absent, due to a relatively shorter 
supraorbital process of the parietal (this character is bilaterally 
variable in the skull of- Platyplecturus). The optic foramen is located 
between the frontal and parietal in Melanophidium and Platy- 
plecturus, but within the frontal in Plecturus, Rhinophis, and 
Uropeltis (Underwood, 1967: 64). In Pseudotyphlops (Fig. 3A) and 
Teretrurus, the optic foramen is a slit-like opening in the posterior 
margin of the frontal. The parietal carries a low sagittal crest in the 
relatively large Pseudotyphlops (Fig. 5C). In the other species with 
smaller skulls, such a sagittal crest is at best very faintly developed 
in the posterior part of the parietal (Fig. 5A-B). In some uropeltines 
such as Rhinophis and Uropeltis (Fig. 5B), the parietals are not 
completely fused in their posterior part. A supratemporal is absent in 
uropeltines, and the quadrate is suspended from the otic capsule in a 
relatively low position. The suprastapedial process of the quadrate is 
very elaborate in uropeltines, and as in Anomochilus (Cundall & 
Rossman 1993), it exceeds the shaft of the quadrate in length. 

In Anomochilus, the anterior end of the edentulous palatine shows 



some elaboration into a broader structure that receives the medial 
(palatine) process of the maxilla in a deep facet (Cundall & Rossman 
1993, Fig. 2B). In uropeltines, the anterior process of the palatine is 
modified to form a broad wing which establishes a broad ventral 
overlap with the posterolateral part of the vomer, and which receives 
the well developed medial process of the maxilla in a deeply 
recessed lateroventrally facing facet (Fig. 2). The infraorbital nerve 
(maxillary division of the trigeminal nerve) pierces the bottom of 
this recessed facet to become the superior alveolar nerve. The 
morphology of the palatine in Anomochilus is intermediate between 
that of Cylindrophis on the one hand, and that of uropeltines on the 
other (Cundall & Rossman 1993). Palatine teeth are absent in 
Anomochilus and uropeltines with the exception of Melanophidium 
wynaudense . The ectopterygoid and pterygoid are reduced in 
uropeltines (and even more so in Anomochilus: Cundall & Rossman 
1993), and the pterygoid is edentulous. 

The para-basisphenoid is relatively broad in uropeltines, gradually 
becoming narrower anteriorly and tapering to a pointed tip between 
the choanal processes of the palatines. The ventral surface of the 
para-basisphenoid is distinctly convex in Pseudotyphlops resulting 
in the formation of ventrolateral ridges. These are at best weakly, or 
only very faintly, developed in other, smaller, species with a para- 
basisphenoid that has a flat or even slightly convex ventral surface. 
Along the lateral edge of the para-basisphenoid the ossified crista 
trabecularis ends behind the anterior margin of the laterally descend- 
ing flange of the parietal in most taxa except for Teretrurus and 
Rhinophis drummondhayi, where it ends at the anterior margin of the 



126 



O. RIEPPEL AND H. ZAHER 








Fig. 4 A-C A, The left maxilla of Cylindrophis maculatus (BMNH 
1930.5.8.48) in lateral and dorsal views; B, The left maxilla of 
Melanophidium wynaudense (BMNH 1930.5.8.124) in lateral and dorsal 
views; C, The left maxilla of Pseudotyphlops philippinus (BMNH 
1978.1092) in lateral and dorsal views. 

laterally descending flange of the parietal. In Rhinophis sanguineus 
and Uropeltis, the crista trabecularis extends further anteriorly and 
terminates below the optic foramen that is located in the frontal. In 
front of the ossified crista trabecularis, the cartilaginous trabecula 
cranii is embedded in all taxa in a deep furrow located between the 
lateral margin of the parasphenoid and the ventrally projecting margin 
of the frontal. Tiny fontanelles may persist along the line of fusion of 
the basisphenoid and basioccipital in Rhinophis and Uropeltis. 



THE OTICO-OCCIPITAL COMPLEX 

The otico-occipital complex is here considered to include the prootic, 
opisthotic-exoccipital, supraoccipital, and basioccipital. These brain- 



case elements show a variable degree of fusion with each other 
among the specimens examined (Fig. 6). All braincase elements 
except the opisthotic and exoccipital remain separate from one 
another in Melanophidium. All braincase elements are fused with 
one another in Plecturus, Pseudotyphlops, Rhinophis and Uropeltis, 
but the basioccipital remains separate from the basisphenoid in 
Teretrurus. The exoccipitals and basioccipital are always fused in 
the occipital condyle. The stalk of the occipital condyle is short in 
Melanophidium, Platyplecturus, and Teretrurus, but distinctly elon- 
gated in the other taxa investigated, such that the depression of the 
basioccipital housing the brainstem is exposed in dorsal view (Fig. 
5). The exoccipitals define the dorsal margin of the foramen mag- 
num, and their posterolateral corners are either deeply notched, or 
perforated by a foramen. 

A laterosphenoid is always present in uropeltines, but while it 
remains a relatively narrow element in Melanophidium (Fig. 6A-B) 
and Pseudotyphlops (Fig. 6C), it becomes distinctly broadened in 
the other taxa investigated. 

The plesiomorphic condition of the posterior opening of the 
Vidian canal and its relation to the facial nerve branches is exempli- 
fied by Melanophidium among uropeltines (Fig. 6A-B). The 
hyomandibular and palatine branches of the facial nerve exit from 
separate foramina opening into an obliquely oriented recess located 
on the prootic closely behind the foramen for the mandibular branch 
of the trigeminal nerve in Melanophidium punctatum (Fig. 6A), and 
incompletely separated from the posterior margin of the mandibular 
branch foramen in Melanophidium wynaudense (Fig. 6B). The 
recess housing the facialis foramina becomes deeper ventrally, as it 
connects with the posterior opening of the Vidian canal that is 
located on the prootic - basisphenoid suture. This condition is 
closely comparable to that in Cylindrophis and Anomochilus (Cundall 
& Rossman 1993), except that the posterior opening of the Vidian 
canal is located more (Anomochilus: Cundall & Rossman 1993, Fig. 
4) or less (Cylindrophis maculatus) below the prootic - basisphe- 
noid suture. In Pseudotyphlops (Fig. 6C) and Rhinophis sanguineus 
(Fig. 6F), the palatine branch of the facial nerve enters directly into 
a canal within the prootic which connects ventrally with the Vidian 
canal, and which opens dorsally within the posteriorly expanded 
recess of the mandibular branch foramen. This prootic canal appears 
to be a modification of the condition observed in Melanophidium by 
the lateral closure of the prootic recess that houses the facialis nerve 
foramina. In Pseudotyphlops and Rhinophis sanguineus the Vidian 
canal retains no separate posterior opening; the internal carotid 
enters directly into the opening of the prootic canal. In all other taxa 
investigated (e.g., Uropeltis, Fig. 6D; Rhinophis drummondhayi. 
Fig. 6E), the palatine branch of the facial nerve enters again a prootic 
canal which is completely separated from the mandibular branch 
foramen however, and which opens anteroventral to the anterior 
corner of the juxtastapedial recess. The internal carotid enters the 
prootic canal on its way to the sella turcica. The anterior opening of 
the Vidian canal lies on the suture between para-basisphenoid and 
parietal in front of the dorsolateral wings of the para-basisphenoid 
(McDowell 1967). 

The juxtastapedial recess is well developed in uropeltines, which 
all except Pseudotyphlops share with Anilius and Cylindrophis the 
presence of a fenestra pseudorotunda (Rieppel 1979a). The shaft of 
the stapes is directed posterolaterally as it connects with the elon- 
gated suprastapedial process of the quadrate via the stylohyal (Rieppel 
1980; see also Wever 1978). As in scolecophidians and basal 
alethinophidians, the juxtastapedial recess is open posteriorly, and 
the jugular foramen is exposed in lateral view (Tchernov etal. 2000; 
Fig. 6). The posteroventral corner of the crista circumfenestralis is 
enlarged to form a gliding surface for the quadrate ramus of the 



SKULL OF UROPELTINAE 



127 




A 1 rnm 



R 1 mm 



01_mnV 



Fig. 5 A-C The otico-occipital region of uropeltine snakes in dorsal views. A, Melanophidium wynaudense (BMNH 1930.5.8.124): B, Uropeltis 
woodmansoni (BMNH 1930.5.8.73); C, Pseudotyphlops philippinus (BMNH 1978.1092). 



op-eo 




Fig. 6 A-F The otico-occipital region of uropeltine snakes in right lateral views. A, Melanophidium punctatum (BMNH 1930.5.8. 119); B, 
Melanophidium wynaudense (BMNH 1930.5.8.124); C, Pseudotyphlops philippinus (BMNH 1978.1092); D, Uropeltis woodmansoni (BMNH 
1930.5.8.73); E, Rhinophis drummondhayi (BMNH 1930.5.8.67-68); F, Rhinophis sanguineus (BMNH 1930.5.8.59). 



128 



O. RIEPPEL AND H. ZAHER 



pterygoid in Melanophidium (Fig. 6A-B ) and Pseudotyphlops (Fig. 
6C), a surface that is 'rounded off to a variable degree in smaller 
species, and reduced in Rhinophis drumrnondhayi (Fig. 6E). By the 
fact that the braincase elements remain separate in Melanophidium, 
it is possible to ascertain that the prootic, opisthotic-exoccipital, and 
basioccipital contribute to this enlarged posteroventral part of the 
crista circumfenestralis. In the plesiomorphic condition, the 
juxtastapedial recess is wide open laterally (Cylindrophis, 
Anomochilus: Cundall & Rossman 1993), and such is also the case 
in Melanophidium (Fig. 6A-B), Platyplecturus and Teretrurus. In 
other uropeltines, the lateral opening of the juxtastapedial recess is 
closed to a narrow slit, most extremely so in Uropeltis (Fig. 6D) and 
Rhinophis (Fig. 6E-F), where the dorsal and ventral lips of the crista 
circumfenestralis closely approach each other, or may even estab- 
lish a restricted contact with each other. Never is the juxtastapedial 
recess fully closed laterally, however, as is the case in Liotyphlops 
(Haas 1964), typhlopids and leptotyphlopids (Rieppel 1979b). The 
jugular foramen is internally subdivided in most uropeltines (except 
in Pseudotyphlops, Teretrurus and Uropeltis), and it is located either 
behind the juxtastapedial recess (plesiomorphic), or in the 
posteroventral corner of the latter (in Platyplecturus, Plecturus, 
Teretrurus and Uropeltis). The exoccipital is pierced by two hy- 
poglossal foramina in Melanophidium punetatum (Fig. 6A), but by 
a single, enlarged hypoglossal foramen in the other taxa investi- 
gated. 



PHYLOGENETIC INTERRELATIONSHIPS 

Recent cladistic analyses hypothesized that Anomochilus is the 
sister-group of uropeltines (Scanlon & Lee 2000; Tchernov et al. 
2000), and Cylindrophis is the sister-group of uropeltines plus 
Anomochilus (Tchernov et al. 2000; see also McDowell 1987; 
Cundall et al. 1993, osteological data only). The addition of soft 
anatomy characters by Cundall et al (1993) resulted in a different 



t J 



/ / i / ./ 



i k 



f / / / / > / / / 



f $ 



/ / / / / / / / / 
/ / / / / it 




Table 1. The data matrix used in the analysis of the phylogenetic 
interrelationships of Uropeltinae. Character definitions are given in 
Appendix I. 

12345 67891 11111 11112 22222 22223 333 
12345 67890 12345 67890 123 



Melanoph. punetatum 


00000 


01000 


00000 


00022 


11111 


11111 


111 


Melanoph. wynaud. 


00100 


00000 

1 

00111 

1 

20111 


00000 


10022 


11111 


11111 


111 


Platyplecturus 


01100 


12011 


10122 


11111 


11121 


111 


Uropeltis 


11111 


12110 


11222 


11111 


11121 


111 


Teretrurus 


10100 


11110 


12001 


10222 


urn 


11121 


111 


Rhinophis drum. 


11111 


mil 


12110 


11222 


mil 


11121 


111 


Rhinophis sang. 


11111 


20111 


11110 


11?22 


urn 


11121 


111 


Plecturus 


11101 


00111 

1 

01111 


12110 


11122 


inn 


11121 


111 


Plseudotyphlops 


11100 


01101 


11222 


inn 


11121 


111 


Anomochilus 


00100 


01000 


0000? 


00011 


mil 


11120 


000 


Cylindrophis 


00000 


00000 

1 

00000 


00000 

1 

00000 


00010 


00001 


11110 


000 


Anilius 


00000 


00000 


00000 


00010 


000 



Fig. 7 The phylogenetic interrelationships of Uropeltinae. See text for 
further discussion. 



cladogram, which still reproduces the monophyly of Alethinophidia 
and Macrostomata respectively, but which shows anilioids (Rieppel 
1977, 1988) to be paraphyletic. The sound transmitting apparatus 
was found to be 'similar' in Typhlops and Rhinophis by Wever 
( 1978: 705), but a sister-group relationship of scolecophidians and 
uropeltines has so far never been recovered through cladistic analy- 
sis of morphological data (Cundall et al. 1993;Kluge, 1991;Scanlon 
and Lee 2000; Tchernov et al. 2000), and it was specifically rejected 
by Cadle et al. (1990; see also Cundall & Rossman 1993). 

Previous work (Tchernov et al. 2000), and the description of the 
uropeltine skull presented above, allows the delimitation of 33 
phylogenetically potentially informative characters (Appendix I 
and Table 1 ) for an analysis of uropeltine interrelationships. Given 
the currently controversial relationships of Anilius and Anomochilus 
relative to uropeltines, these two taxa together with Cylindrophis 
were used as paraphyletic outgroup in the analysis of uropeltine in- 
group relationships (characters 25 through 28 are uninformative 
using this rooting procedure, and were ignored in the analysis). The 
analysis was performed using PAUP version 3.1.1. (Swofford 1991. 
Swofford & Begle 1993). All multistate characters were unordered, 
and the branch-and-bound search option was implemented. Character 
optimization is based on the DELTRAN routine. 

A single most parsimonious tree was obtained (TL = 49; CI = 
0.796; RI = 0.877) with fully resolved uropeltine relationships. 
Given the scarcity of characters, it is not surprising that some nodes 
among uropeltines are rather poorly supported (with the minimal 
decay index or Bremer support index [Bremer 1 988] : + 1 ). Neverthe- 
less, the tree (Fig. 7) suggests some interesting preliminary results. 

The basal position of Melanophidium among uropeltines was 
expected (Rieppel 1977; McDowell 1987), and is reproduced here. 
However, there is a signal for paraphyly of the genus Melanophidium. 
Melanophidium wynaudense appears to be more closely related to 
other uropeltines than it is to Melanophidium punetatum (decay 
index: +1) on the basis of the presence of a single (enlarged) 
hypoglossal foramen behind the jugular foramen ( 16[1] ; ci=l). 
Evidently, the monophyly of the genus Melanophidium must be 
tested by the addition of other characters, including soft anatomy, 
because the result obtained here may be nothing more than the 
reflection of the fact that as coded, Melanophidium punetatum is 
plesiomorphic relative to all other uropeltines in all characters that 



SKULL OF UROPELTINAE 



129 



are informative for the analysis of uropeltine interrelationships 
(unfortunately. Melanophidium was not included in the analysis 
performed by Cadle et al. [1990]). 

The genera Teret runts. Platyplecturus, Pseudotyphlops, Plecturus, 
Uropeltis. and Rhinophis form a monophyletic clade that is very 
strongly supported on the basis of 7 characters (decay index: +6). 
This represents a strong corroboration of the basal position of the 
genus Melanophidium (unequivocal synapomorphies are designated 
with an asterisk): *8(1), supraoccipital fused to opisthotic - 
exoccipital; *9(1) prootic fused to opisthotic - exoccipital; 11(1) 
laterosphenoid broad (reversal implied); 12(2) palatine branch of 
facial nerve enclosed in prootic canal which is separate from man- 
dibular branch foramen; 15(1) jugular foramen single (reversals 
implied); 18(2) posteroventral process of dentary absent (reversal 
implied); 29(2) gliding surface for pterygoid posteroventral to 
juxtastapedial recess 'rounded off (reversal implied). The 
monophyly of all uropeltines except Melanophidium is the most 
strongly supported clade on the basis of this data set. 

Within that clade, Platyplecturus is the sister-taxon of a clade that 
includes (Pseudotyphlops (Plecturus (Rhinophis, Uropeltis))) on 
the basis of 2 characters (decay index: +1): *2(1) nasals narrow 
anteriorly, gradually tapering to pointed tip; *10(1) basioccipital 
fused to basisphenoid. Pseudotyphlops is the sister-taxon of a clade 
that includes (Plecturus (Rhinophis, Uropeltis)) on the basis of three 
characters (decay index: + 1 ): 1(1) transverse process of the premax- 
illa points posterolaterally, and meets the straight maxilla in a 
shizarthrosis (convergent in Teretrurus); * 1 3( 1 ) narrow lateral open- 
ing of the juxtastapedial recess; * 17(1) stalk of the occipital condyle 
elongated. Plecturus shares with the (Rhinophis. Uropeltis) - clade 
three characters (decay index: +2): 5( 1 ) optic foramen fully enclosed 
by frontal; 14(1) jugular foramen recessed within posteroventral 
corner of juxtastapedial recess; 15(0), jugular foramen internally 
subdivided (reversal). 

The clade that includes Rhinophis and Uropeltis (decay index: 
+ 1) is diagnosed by a well-defined buttressing contact between the 
processus medialis anterior of the maxilla and an anterior lateral 
process of the vomer (*4 [1]). Interestingly, there is a signal for the 
paraphyly of the genus Rhinophis, because Rhinophis sanguineus 
appears to be more closely related to Uropeltis than to Rhinophis 
drummondhayi on the basis of two characters (decay index: +1): 
6(2) crista trabecularis ends in front of lateral fronto-parietal suture; 
7(0) supraorbital process of parietal does not contact prefrontal 
(reversal). 



1990) is not supported here. By contrast, the possible paraphyly of 
the genus Rhinophis indicated by molecular data (Cadle et al. 1990) 
is also found here, although far less species were included in the 
morphological analysis. 

Pseudotyphlops is larger that all other uropeltines included in the 
analysis, and it shows characters of cranial anatomy that appear in 
outgroup taxa such as Anilius and Cylindrophis, but not in other 
uropeltines. In the adult skull of Anilius and Cylindrophis, the part of 
the para-basisphenoid located behind the optic foramen has a con- 
cave ventral surface, which results in the formation of distinct lateral 
ventral ridges (Tchernov etal. 2000). This character is also observed 
in the relatively large skull of adult Pseudotyphlops. In the much 
smaller skull of other uropeltines, the ventral surface of the para- 
basisphenoid is at best very weakly concave, flat, or even slightly 
convex, and ventral lateral ridges are very faintly indicated 
(Melanophidium. Platyplecturus. Plecturus. Uropeltis. Rhinophis), 
or absent (Teretrurus; also in Anomochilus: Cundall & Rossman 
1993). The same observation relates to the presence of a sagittal 
ridge on the parietal, well expressed in adult Anilius, Cylindrophis, 
and in Pseudotyphlops among uropeltines, much reduced and re- 
stricted to the posterior part of the parietal or absent in Anomochilus 
and smaller uropeltines. Given its relative size, and the presence of 
relatively plesiomorphic features in the skull, a basal position of 
Pseudotyphlops relative to other uropeltines might have been 
expected, but was not corroborated by cladistic analysis, although 
the genus is still outside the (Plecturus (Rhinophis. Uropeltis)) 
clade. 

The morphological transformation that is implied in the descrip- 
tion of the Vidian canal in uropeltines (and in its coding; the 
character was used unordered) is also contradicted by the cladistic 
analysis discussed above. The description suggests that the indi- 
vidualization of the prootic canal (which receives the palatine 
branch of the facial nerve and into which enters the internal carotid) 
follows its formation in association with the recess of the mandibu- 
lar branch foramen, the cladistic analysis suggests otherwise. The 
incorporation of the opening of the prootic canal into the recess of 
the mandibular branch foramen is a secondary development that 
occurred convergently in Pseudotyphlops and Rhinophis sanguineus. 



Acknowledgements. We would like to thank E.N. Arnold and C. 
McCarthy (BMNH), as well as Harold Voris and Alan Resetar (FMNH) for 
permission to study the collections under their care. D. Cundall and two 
anonymous reviewers offered helpful criticism on earlier drafts of this paper. 



DISCUSSION AND CONCLUSIONS 

The monophyly of Uropeltinae has not previously been questioned 
(Rieppel 1977; Cadle et al. 1990; Cundall et al. 1993;Scanlon&Lee 
2000; Tchernov et al. 2000) and is here corroborated by six un- 
equivocal synapomorphies (decay index: +3): *19(2), exoccipitals 
and basioccipital fused in occipital condyle; *20(2), anterior denti- 
gerous process of palatine modified into expanded lamina; *30(1), 
occipital condyle modified as described by Williams (1959) and 
Hoffstetter & Gasc ( 1 969); *3 1 ( 1 ), the superior alveolar nerve canal 
in the maxilla is open dorsally ; *32( 1 ), frontals at least twice as long 
as broad; *33( 1 ), supratemporal absent. 

The phylogenetic analysis of the interrelationships among 
Uropeltinae corroborates the hypothesized basal position of the 
genus Melanophidium, the latter possibly paraphyletic. The clade 
comprising Teretrurus and the Indian species of Uropeltis that is 
consistently obtained on the basis of allozyme data (Cadle et al. 



REFERENCES 



Baumeister, L. 1908. Beitrage zur Anatomie und Physiologie der Rhinophiden. 

Zoologische JahrbUcher, Abteilung fur Anatomie und Ontogenie der Here, 26: 423- 

526. 
Bellairs, A.d'A. & Kamal, A. 1981. The chondrocranium and the development of the 

skull in Recent reptiles, pp 2-263. In: Gans. C. & Parsons, T.S. (eds). Biology oj the 

Reptilia. Vol. 2. Academic Press, London. 
Bremer, K. 1988. The limits of amino acid sequence data in angiosperm phylogenetic 

reconstruction. Evolution 42: 795-803. 
Cadle, J.E., Dessauer, H.C., Gans, C. & Gartside, D.F. 1990. Phylogenetic relation- 
ships and molecular evolution in uropeltid snakes (Serpentes: Uropeltidae): allozymes 

and albumin immunology. Biological Journal of the Linnean Society 40: 293-320. 
Cundall. D. & Greene, H.W. 2000. Feeding in snakes, pp. 293-333. In Schwenk, K. 

(ed). Feeding: Form, Function, and Evolution in Tetrapod Vertebrates. Academic 

Press. San Diego. 
& Rossman, D.A. 1993. Cephalic anatomy of the rare Indonesian snake 

Anomochilus weberi. Zoological Journal of the Linnean Society 109: 235-273. 
Wallach, V. & Rossman, D.A. 1993. The systematic relationships of the snake 

genus Anomochilus. Zoological Journal of the Linnean Society 109: 275-299. 



130 



O. RIEPPEL AND H. ZAHER 



Frazzetta, T.H. 1966. Studies on the morphology and function of the skull in the 
Boidae (Serpentes). Part 2. Morphology and function of the jaw apparatus in Python 
sebae and Python molurus. Journal of Morphology 118: 217-296. 

Haas, G. 1930. Uber die Kaumuskulatur und die Schadelmechanik einiger 
Wuhlschlangen. Zoologische Jahrbiicher, Abteilung fur Anatomie und Ontogenie 
der Tiere 52: 1-94. 

Hoffstetter, R. & Gasc, J.-P. 1969. Vertebrae and ribs of Modern reptiles, pp. 201-310. 
In: Gans, C. Parsons, T.S. & Bellairs, A.d'A. (eds). Biology of the Reptilia Vol. 1. 
Academic Press, London. 

Kluge, A.G. 1991. Boine snake phylogeny and research cycles. Miscellaneous Publi- 
cations Museum of Zoology, University of Michigan, 178: 1-58. 

McDowell, S.B. 1967. Osteology of the Typhlopidae and Leptotyphlopidae: a critical 
review. Copeia 1967: 686-692. 

1987. Systematics. pp 3-50. In: Seigel, R.A., Collins, J.T. & Novak, S.S. (eds). 

Snakes. Ecology and Evolutionary Biology. Macmillan Publ., New York. 

Nopcsa, F.v. 1923. Die Familien der Reptilien. Fortschritte der Geologie und 
Paldontologie 2: 1-210. 

Rieppel, O. 1977. Studies on the skull of the Henophidia (Reptilia, Serpentes). Journal 
of Zoology. London 181: 145-173. 

1978. The evolution of the naso-frontal joint in snakes and its bearing on snake 

origins. Zeitschrift fUr zoologische Systematik und Evolutionsforschung 16: 14—27. 

1979a. The evolution of the basicranium in the Henophidia (Reptilia, Serpentes). 

Zoological Journal of the Linnean Society 66: 41 1-431. 

1979b. The braincase of Typhlops and Leptotyphlops (Reptilia. Serpentes). 

Zoological Journal of the Linnean Society, 65: 161-176. 

1980. The sound transmitting apparatus of primitive snakes and its phylogenetic 

significance. Zoomorphology 96: 45-62. 
1983. A comparison of the skull of Lanthanotus borneensis (Reptilia: Varanoidea) 

with the skull of primitive snakes. Zeitschrift fur zoologische Systematik und 

Evolutionsforschung 21: 142-153. 

1988. A review of the origin of snakes. Evolutionary Biology 22: 37-130. 

& Zaher, H. 2000. The intramandibular joint in squamates. and the phylogenetic 

relationships of the fossil snake Pachyrhachis problematicus Haas. Fieldiana (Geol- 
ogy), n.s. 43: 1-69. 
Scanlon, J.D. & Lee, M.S.Y. 2000. The Pleistocene serpent Wonambi and the early 

evolution of snakes. Nature 403: 416^20. 
Swofford, D.L. 1990. PAUP - Phylogenetic Analysis Using Parsimony, Version 3.0. 

Illinois Natural History Survey. Champaign. 
& Begle, D.P. 1993. PAUP - Phylogenetic Analysis Using Parsimony, Version 

J. I . Laboratory of Molecular Systematics, Smithsonian Institution, Washington DC. 
Tchernov, E„ Rieppel, O., Zaher, H., Polcyn, M.J. & Jacobs, L.J. 2000. A new fossil 

snake with limbs. Science 287: 2010-2012. 
Underwood, G. 1967. A Contribution to the Classification of Snakes. British Museum 

(Natural History) Publication No. 653 Trustees of the British Museum (Natural 

History), London. 179pp. 
Wever, E.G. 1978. The Reptile Ear. Princeton University Press, Princeton. 1024pp. 
Williams, E.E. 1959. The occipito- vertebral joint in the burrowing snakes of the family 

Uropeltidae. Breviora 106: 1-10. 

Appendix I List of Characters used in the 
phylogenetic analysis 

1. Anterior tip of maxilla turned medially, closely approaching or 
touching transverse process of premaxilla (0); anterior tip of 
maxilla straight, contact with premaxilla shizarthrotic (1). 

2. Nasals relatively broad anteriorly, notched (0); nasals gradually 
tapering to pointed tip anteriorly ( 1 ). 

3. Teeth on palatine present (0), absent (1). 

4. A distinct and well defined buttressing contact between the 
processus medialis anterior of the maxilla and an anterior lateral 
process of the vomer is absent (0), or present (1). 

5. Parietal enters optic foramen (0), optic foramen fully enclosed 
entirely within frontal (1). 

6. Crista trabecularis ends behind the (lateral) fronto-parietal su- 
ture (0), at the (lateral) fronto-parietal suture (1), in front of the 
(lateral) fronto-parietal suture (2). 

7. Supraorbital process of parietal does not (0), or does ( 1 ) participate 
in suspension of prefrontal (contacts prefrontal above the orbit). 



8. Supraoccipital separate (0), or fused ( 1 ). 

9. Prootic and opisthotic-exoccipital separate (0), or fused (1). 

10. Basisphenoid - basioccipital separate (0), or fused (1). 

11. Laterosphenoid narrow (0), broad (1). 

12. Facial nerve branches open into a recess behind the mandibular 
branch foramen which connects with the posterior opening of 
the Vidian canal (0); facial nerve branches open into a prootic 
canal which opens within the recession of the mandibular branch 
foramen and connects with the posterior opening of the Vidian 
canal (1); facial nerve branches open into a prootic canal which 
opens behind the mandibular branch foramen and connects with 
the posterior opening of the Vidian canal (2). 

13. Juxtastapedial recess wide open laterally (0; fenestra pseudo- 
rotunda may be exposed in lateral view), distinctly restricted by 
approximation of dorsal and ventral margin (1; fenestra 
pseudorotunda never exposed in lateral view). 

14. Jugular foramen behind juxtastapedial recess (0), recessed within 
juxtastapedial recess (1). 

15. Jugular foramen internally subdivided (0), single (1). 

16. More than one hypoglossal foramina (0), single but enlarged 
hypoglossal foramen ( 1 ). 

17. Stalk of occipital condyle short, depression in basioccipital for 
brainstem not visible in dorsal view (0), stalk of occipital 
condyle elongate, depression in basioccipital for brainstem 
visible in dorsal view (1). 

18. Posteroventral process of dentary distinct (0), reduced (1), 
absent (2). 

19. Exoccipitals not in contact dorsal to basioccipital in occipital 
condyle (0); exoccipitals in contact dorsal to basioccipital in 
occipital condyle (1); exoccipitals and basioccipital fused in 
occipital condyle (2). 

20. Anterior dentigerous process of palatine slender and straight (0), 
broadened anteriorly ( 1 ), modified into expanded lamina (2). 

21. Pterygoid teeth present (0), absent (1) (Tchernov et al. 2000). 

22. Suprastapedial process of stapes is not (0), or is (1) distinctly 
longer than shaft of stapes (Tchernov et al. 2000). 

23. Quadrate suspension close to dorsal margin of otic capsule (0), 
shifted anteroventrally on otic capsule (1) (Tchernov et al. 
2000). 

24. Retroarticular process unmodified (0), wrapping around poste- 
rior aspect of mandibular condyle of quadrate ( 1 ) (Cundall et al. 
1993). 

25. Premaxillary teeth present (0), absent (1). 

26. Contact between premaxilla and vomer overlapping (0), or in 
well defined recess (1). 

27. Preorbital ridge on frontal (Frazzetta 1 966) does not (0), does ( 1 ) 
project beyond anterior margin of dorsally exposed surface of 
frontal. 

28. Interchoanal process of parasphenoid absent (0), present (1). 

29. Posteroventral part of crista circumfenestralis does not (0), or 
does (1) form a distinctly enlarged gliding surface for the 
quadrate ramus of the pterygoid, or this gliding surface is 
present but "rounded off (2). 

30. Occipital condyle is not (0), or is (1) modified as described by 
Williams (1959) and Hoffstetter & Gasc (1969). 

31. The superior alveolar nerve canal in the maxilla is closed (0), or 
open ( 1 ) dorsally. 

32. Frontals are not (0), are (1) at least twice as long as broad. 

33. Supratemporal present (0), absent ( 1 ). 



XX fiS7/30,/ 



fii///. mtf. «/.s7. Mus. bond. (Zool.) 68(2): 131-142 



Issued 28 November 2002 



The Cretaceous marine squamate Mesoleptos 
and the origin of snakes 

MICHAEL S. Y. LEE AND JOHN D. SCANLON 

Department of Palaeontology, The South Australian Museum, North Terrace, Adelaide SA 5000, Australia, e- 

mail Lee. Mike® saugov.sa.gov.au 

Department of Environmental Biology, The University of Adelaide, Adelaide SA 5005, Australia 



SYNOPSIS. The poorly known marine squamate Mesoleptos is reassessed based on two previously known specimens and a 
newly referred specimen. The three specimens of Mesoleptos zendrinii share unique characters such as long, posteriorly tapering 
centra and distally straight but non-pachyostotic ribs. Mesoleptos had a narrow neck (and presumably small head), long laterally 
compressed body, and small fore- and hindlimbs. Phylogenetic analysis suggests that Mesoleptos is the nearest relative of snakes 
this phylogenetic position is consistent with its morphology being intermediate between typical marine squamates (e.g 
mosasauroids) and primitive marine snakes (pachyophiids). However, this interpretation remains tentative because Mesoleptos 
is very poorly known, and many of the characters uniting it with mosasauroids and primitive snakes are correlates of marine habits 
and/or limb reduction. 



INTRODUCTION 



Whereas sea snakes (Laticaudinae and Hydrophiinae) and marine 
iguanas (Amblyrhyncluts) are the only truly marine squamates living 
today, there was a more diverse and very different radiation of such 
forms during the Cretaceous. These extinct marine squamates 
included the large monitor-like aigialosaurs and mosasaurs, the 
small, long-necked dolichosaurs, and the medium-sized limbed 
snakes Pachyrhachis, Pachyophis, and Haasiophis. These forms 
were suggested by workers in the late nineteenth and early twentieth 
centuries to be closely related to each other and to modern snakes 
(e.g. Cope, 1869; Boulenger, 1891; Gorjanovic-Kramberger, 1892; 
Nopcsa, 1908, 1923), a view which has been supported by some 
recent phylogenetic analyses (e.g. Scanlon, 1996; Caldwell 1999; 
Lee and Scanlon 2002; but see Tchernov et al. 2000; Rieppel and 
Zaher 2000). 

One poorly known form that has been associated with this radia- 
tion is Mesoleptos zendrinii (Cornalia and Chiozza, 1852; 
Gorjanovic-Kramberger, 1 892; Calligaris, 1988). M. zendrinii was a 
marine squamate with a rather elongated body, long ribs, and well- 
developed but rather small hindlimbs. It has been repeatedly 
associated with other contemporary marine squamates, largely on 
the basis of common habitat rather than any detailed analysis of 
morphology. Cornalia and Chiozza (1852) suggested affinities with 
i Raphiosaurus\ based on a specimen (BMNH R32268) figured 
under this name by Owen ( 1 842) but later referred to Dolichosaurus 
(Owen 1850a, 1851). Subsequent workers have commented on 
errors in the original description, though a full redescription of the 
type specimen has not appeared. Gorjanovic-Kramberger (1892) 
referred an additional specimen to Mesoleptos cf. zendrinii, dis- 
cussed below, and referred this genus to the Varanidae, although 
acknowledging that it differed from other varanids in being highly 
aquatic. Nopcsa (1903) referred it tentatively to Aigialosauridae, 
and suggested that the moderate elongation of the trunk region 
relative to other known aigialosaurs was analogous to the independ- 
ent elongation of the body in some mosasaurs such as Clidastes. 
Later, Nopcsa (1923) compared M. zendrinii with Eidolosaunts 
trauthi, including both in a subfamily Mesoleptinae within his 
broadly conceived Dolichosauridae (Mesoleptinae, Aigialosaurinae, 
Dolichosaurinae). He regarded the Mesoleptinae as intermediate 



between two main lineages, one consisting of the Aigialosaurinae 
plus their probable descendants the Mosasauridae, and the other 
consisting of the Dolichosaurinae plus their probable nearest rela- 
tives - though not direct descendants - the snakes. Nopcsa's ( 1 903, 
1 923) classifications still represent the most complete discussion of 
these forms to date, and are summarised by Calligaris (1988). 
However, no unambiguous derived characters have been proposed 
linking Mesoleptos with any of the other marine groups or with 
snakes, and these interpretations need to be critically examined. 

Here, we identify a new specimen of Mesoleptos, compare it to 
previously known specimens, and use the combined material to infer 
the phylogenetic relationships and palaeoecology of Mesoleptos. 
Mesoleptos emerges as on the stem lineage leading to snakes, lying 
phylogenetically between marine lizards (mosasauroids, 
dolichosaurs, Adriosaurus) and primitive limbed snakes 
(Pachyrhachis. Pachyophis, Haasiophis). Garth Underwood's ear- 
liest research interests included the origin and evolution of snakes, 
and he has contributed to possibly the two most influential papers on 
this topic (Bellairs and Underwood 1951; Underwood 1967). The 
current paper is thus a small contribution to a field of inquiry that 
Garth Underwood helped establish. 

Institutional abbreviations 

HUJ PAL, Hebrew University of Jerusalem Palaeontological Col- 
lection; MCSNT, Museo Civico di Storia Naturale di Trieste; MNHN, 
Musee Nationale d'Histoire Naturelle, Paris; SAM, South Austral- 
ian Museum. 



DESCRIPTION OF NEW SPECIMEN 

Material and horizon 

The specimen consists of part and counterpart, but all morphological 
information is preserved on the part (Fig. 1A). Anterior vertebral 
column, ribs, shoulder girdle, and partial forelimbs. Locality: 'Ein 
Jabrud (Ain Yabrud), 7 km north-east of Ramallah (West Bank, 
Palestine) and 20 km north of Jerusalem. Stratigraphic horizon: Bet- 
Meir Formation (Lower Cenomanian; earliest Upper Cretaceous). 
Catalogued as HUJ-PAL EJ699. 



© The Natural History Museum, 2002 



132 



M.S.Y. LEE AND J.D. SCANLON 









p.vert 




Vertebrae 

An articulated series of thirteen vertebrae (here referred to as 
vertebrae 1-13) is preserved, along with an isolated element on the 
lower left (vertebra 14). All vertebrae are exposed ventrally only; 
the surfaces of vertebrae 1-7 are weathered, while that of vertebra 1 1 
is broken. The series 1-13 represents the anterior presacral part of 
the column. Vertebra 1 , the anteriormost, is the smallest; size then 
increases gradually along the series such that the last is approxi- 
mately twice the dimension of the first. The cervical-dorsal boundary 
cannot be precisely determined because the cartilaginous sternal 
contacts are not preserved. However, in typical lizards (anonymous 
referee, pers. comm.), the cervical-dorsal boundary lies slightly 
behind an abrupt increase in rib length. There is an abrupt change in 
the size and shape of the ribs between preserved vertebrae 5 and 6 
(see below), suggesting the cervical-dorsal boundary was slightly 
behind this region, perhaps between vertebrae 7 and 8. Both shoulder 
girdles, however, are preserved around the level of vertebra 5, 
suggesting a slightly more anterior cervical-dorsal boundary. 

The centra are all elongate, the length being approximately three 
times the width across the middle of each vertebra. They narrow 
sharply behind the transverse processes, and then more gradually 
posteriorly. All centra are procoelous; the anterior cotyle is deeply 
concave and the posterior condyle strongly convex. The articular 
surfaces of the condyles face posteriorly; part of the surface is 
sometimes exposed in ventral view, so they were at most only 
slightly inclined dorsally. 

Subcentral foramina are visible on the ventral surface of most 
vertebrae: two are present on vertebrae 6 to 9, and one is present on 
vertebrae 10 and 12. They were presumably present on the other 
vertebrae but are not visible due to weathering and/or damage. 
Where two foramina are present on a single vertebrae, they are never 
bilaterally symmetrical and are often both on the same side of the 
midline. 

A sagittal keel, extending along the posterior half of the centrum, 
is present on vertebrae 1 to 7. The keel terminates posteriorly in a 
prominent knob-shaped hypapophysis, which is, however, partly 
weathered away on all except vertebrae 6 and 7. The keels and 
(where preserved) the hypapophyses are more prominent on the 
anteriormost vertebrae and gradually decrease in size posteriorly. 
On vertebra 8, there is no keel. A weak hypapophysis may have been 
present, but this cannot be confirmed due to breakage. Both the keel 
and hypapophysis are absent from vertebrae 9 to 13, and the ventral 
surface is completely smooth. 

A pair of transverse processes extend laterally from the anterior 
end of each centrum. These processes extend proportionally further 
laterally in the more posterior vertebrae: the diameter across the 
transverse processes is slightly less than the length of the centrum in 
the anteriormost vertebra, but slightly more in the posteriormost 
vertebra (Table 1). Most of the tranverse processes on the anterior 
vertebrae are weathered ventrally, but at least one is complete on 
most of the posterior vertebrae. The articular surfaces of the proc- 
esses are not fully exposed, but appear to have been single based on 
the morphology of the proximal ends of the ribs. 

The isolated vertebra '14' does not fit onto either end of the 



Fig. 1 (A) Photograph of the third known individual of Mesoleptos (HUJ- 
PAL 699). (B) Specimen drawing. The anterior end of the specimen is to 
the top right. Unstippled areas represent areas repesent broken bone. 
Scale bar = 2cm. Abbreviations: cor, coracoid; sea, scapula; cla, 
clavicle; hum. humerus; ep, epiphyseal ossification; vl first 
(anteriormost) preserved vertebra; p.vert, isolated posterior vertebra: r5. 
rib of fifth preserved vertebra: hyp. hypapophysis. 



MESOLEPTOS AND THE ORIGIN OF SNAKES 



133 



Table 1. Measurements of'HUJ-PAL EJ699: midline length between rims 
of cotyle and condyle; width across transverse processes; straight-line 
length of rib. The vertebrae are numbered from the first preserved 
centrum. 



Vertebra no. 


Centrum length 


Greatest width 


Rib length 


1 


22 


16 


_ 


2 


20.5 


19 


- 


3 


22 


20 


- 


4 


- 


- 


19.5 


5 


19 


20 


20 


6 


20 


20.5 


41 


7 


21 


20 


49 


8 


(16+) 


23 


49 


9 


22 


23 


59 


10 


23.5 


24 


73 


II 


(23+) 


(24+) 


89 


12 


25 


31 


(92+) 


13 


26 


39 


121 


- 


- 


- 


132 


- 


- 


- 


139 


14 


23 


34.5 


- 



articulated series. It is too large to fit next to vertebra 1, and 
furthermore could not be a cervical as it lacks the mid-ventral keel 
and hypapophysis. However, it is too small to fit next to vertebra 1 3. 
As in most squamates, after reaching maximum size (at or past 
vertebra 13), the centra must have again gradually decreased in size 
towards the posterior end of the dorsal region. The isolated vertebra 
appears to belong to this region. Its surface is worn in a manner that 
suggests there were laterally paired ventral mounds or processes 
defining a median longitudinal trough on the posterior part of the 
centrum. 

Ribs 

Ribs are preserved in association with vertebrae 4 to 1 3. Only the left 
rib (right in ventral view) of vertebra 4 is preserved. Both ribs are 
preserved in association with vertebrae 5 to 8. Only the left ribs are 
associated with vertebrae 9 to 11. Both ribs are associated with 
vertebrae 12 and 13, but the right ribs are displaced so that they 
overlie the left ribs and point anteriorly. Three additional ribs 
belonging to the next three (missing) dorsal vertebrae are also 
preserved; these are presumably right ribs based on their similar 
orientation to the right ribs of the last two preserved vertebrae. 

The anteriormost preserved rib is associated with the 4th vertebra. 
It is short (only as long as the centrum) and smoothly curved. The 
shaft is oval in cross-section and uniformly thick throughout its 
length. Slightly longer ribs, of similar shape, are associated with the 
5th vertebra. The next pair of ribs, associated with the 6th vertebra, 
are much longer and quite different in shape. The distal end of the 
left rib (right in ventral view) is weathered away; the right rib is 
complete and its proximal half is smoothly curved, but the distal half 
is nearly straight. The more posterior ribs are similar in shape, 
except that the curved proximal portion occupies progressively less 
and less of the shaft. By vertebra 13, the curved portion only 
occupies the proximal one-fifth of the shaft. 

The articular surfaces are visible on the left ribs associated with 
vertebrae 7, 8, 10 and 13, and on the second of the three isolated ribs. 
The ribs are all single-headed. The anterior ribs are flared at the 
proximal end and then nearly uniformly thick thoughout their 
length, while more posterior ones have a distinct neck proximally 
before becoming thickened in the region of greatest curvature, then 
gradually tapering distally in the straight part of the shaft. The distal 
ends are truncated squarely where they joined the costal cartilages. 



which are not preserved. 

Approximate measurements of the vertebrae and ribs (Table 1) 
show a more or less steady increase in dimensions from vertebra 1 to 
1 3, continued in the ribs belonging to the next two missing vertebrae 
(both ends of the last known rib are incomplete or obscured and its 
length is therefore not measurable). As noted above, the posteriormost 
preserved ribs cannot belong to vertebra 14, which is from the 
posterior trunk (abdominal) region. 

Shoulder girdle and forelimb 

Both scapulocoracoids are preserved in medial view. The right is 
complete except for the dorsal scapular blade, while the left is partly 
covered by a rib and is missing the distal (anterior) end of the 
procoracoid process. A curved strip of bone adjacent to the left 
scapulocoracoid is probably the left clavicle. The left humerus is 
preserved in proximal dorsal view. All appendicular elements are 
very small in proportion to the axial elements. 

The scapula is a simple, rectangular plate; the scapular blade is 
short. Its anterior margin is weakly concave; a scapulocoracoid 
emargination was thus present. The coracoid is single and bears two 
processes, and two emarginations. The more dorsal process is much 
longer and extends anterodorsally. forming the ventral margin of the 
scapulocoracoid foramen and the dorsal margin of the coracoid 
emargination that represents the anterior coracoid foramen. The 
ventral process is shorter and expanded distally. It forms the ventral 
border of the anterior coracoid foramen and the dorsal border of the 
emargination representing the posterior coracoid foramen. The ven- 
tral margin of the coracoid is smoothly convex, and the posterior 
margin is drawn out into a posteroventral spur. The probable clavicle 
is a tiny curved rod, tapered at each end. There is no ventromedial 
expansion or foramen. The humerus is relatively large compared to 
the shoulder elements, though still small compared to the axial 
elements. The proximal end is expanded and flattened. The entire 
articular surface is occupied by a large, semilunar epiphysis which 
caps the humerus. The distal end of the humerus is weathered. 



COMPARISONS WITH SIMILAR TAXA 

The specimen is clearly a squamate, as it possesses all the 
synapomorphies of squamates (Estes etal., 1988) for which it can be 
coded: single-headed ribs, cervical vertebrae with hypapophyses, 
procoelous vertebrae, presence of anterior coracoid emargination. 
Admittedly, these are relatively few because the specimen is very 
incomplete, but still sufficient to make a firm identification. Among 
squamates, it is clearly different from most groups in possessing 
distally straight ribs. The only taxa that possess such ribs are 
Mesoleptos, Adriosaiirus,Acteosaurus, and various groups of aquatic 
snakes. The specimen here is compared to these forms, and to some 
other superficially similar taxa to which it might be related. 

Mesoleptos zendrinii 

HUJ-PAL EJ699 is extremely similar to Mesoleptos, which is 
known from two specimens. The type of Mesoloptos zendrinii, from 
the Upper Cretaceous of Comen, Slovenia, is an articulated series of 
dorsal, sacral and anterior caudal vertebrae with ribs and a partial 
hindlimb. The specimen has been illustrated as a lithographic plate 
(Cornalia and Chiozza, 1852: pi. 3) and an interpretive line drawing 
(Calligaris, 1988: fig. 2). Cornalia (in Cornalia and Chiozza, 1852) 
considered the specimen to be exposed in dorsal view, while 
Gorjanovic-Kramberger ( 1 892) maintained it was exposed ventrally, 



134 



M.S.Y. LEE AND J.D. SCANLON 



Table 2. Measurements of Mesoleptos zertdrinii holotype (based on 
Cornalia and Chiozza, 1852: pi. 3), for comparison with data in Table 1. 
The vertebrae are numbered from the first preserved rib. 



Vertebra no. 


Centrum length 


Greatest width 


Rib length 


1 
2 
2 


- 


- 


45 


_ 


_ 


75 


4 


12 


- 


90 


5 


13 


- 


116 


6 


12 


- 


120 


7 


12 


18 


125 


8 


12 


20 


119+ 


9 


20* 


23 


124+ 


10 


15 


24 


120+ 


11 


15 


25 


120+ 


12 


15 


22 


114 



* there may be inaccuracies with the outlines of some vertebrae in the original 
figure, or this anomalous high value could reflect longitudinal separation of two 
adjacent vertebrae during partial disarticulation of the skeleton before fossilisation. 



but in any case most of the vertebrae are bisected by the broken 
surface of the slab and are thus seen as cross-sections at various 
levels. The intervertebral articulations are not clearly exposed, and 
Cornalia found no indication that the vertebrae were procoelous, 
though Gorjanovic-Kramberger (1892) and later authors assumed 
that they must have been similar to the specimen in the Novak 
collection (discussed below). The type specimen could not be 
located in recent times: Calligaris (1988) was unable to confirm it 
was still in the Museo Civico di Storia Naturale de Milano (Milan). 
The most anterior parts preserved of the type are strongly curved 
ribs which probably contacted the sternum, and the first vertebral 
fragments are associated with the fourth visible rib. Some small 
elements and fragments visible between the anterior ribs may include 
parts of the shoulder girdle and/or forelimb. Apart from the first few, 
the ribs are weakly curved proximally and nearly straight for the 
distal two-thirds of their length. The ribs are widest at the proximal 
articulation and are otherwise slender, with no trace of thickening 
(pachyostosis) more distally. Ribs in the posterior half of the trunk 
are displaced to point anteriorly, corresponding to bloating and 
maceration of the carcass proceeding most rapidly in the area of the 
viscera, and the most posterior ribs are either lost or not exposed. 
The outlines of the first 12 preserved vertebrae are nearly triangular, 
indicating that they are split horizontally through the middle or 
lower part of the centrum. From about the 1 3th preserved vertebra 
the outlines of the trunk vertebrae are expanded posteriorly as well 
as anteriorly and the neural canal is exposed, indicating a more 
dorsal position of the break; after the 22nd there is not much visible 
of the vertebral centra themselves. Prominent transverse processes 
are visible on vertebrae 24-27, and transverse grooves on the 24th 
and 26th vertebrae resemble lymph channels seen on the ventral 
surface of the sacral and anterior caudal vertebrae in Varanus, 
suggesting that the skeleton is exposed ventrally, and that the 24th 
and 25th preserved vertebrae are the sacrals. After the first two 
caudals (26-27), represented by broad transverse processes of one 
side, there are indeterminate fragments of two more vertebrae, then 
indications of four vertebrae in lateral view showing elongate, near- 
vertical chevrons and a tall but antero-posteriorly narrow, slightly 
back-sloping neural spine. Traces of longitudinal elements under the 
transverse processes of the 25th-26th probably represent the ilium, 
slightly displaced posteriorly, medially and (if the orientation is 
correct) dorsally from its natural position. The femur is level with 
the probable sacrals; the tibia and fibula are articulated, but incom- 
plete distally. 



The two referred specimens consist of HUJ-PAL EJ699 and 
another specimen in the Museo Civico di Storia Naturale. Trieste 
(MCSNT 9962: locality and other collection details undetermined). 
The latter consists of a shorter but similar section of the skeleton to 
that in the type, exposed dorsally (Calligaris. 1988). Comparisons of 
the vertebrae are difficult due to the different parts and orientations 
of the skeleton in the different specimens, but all three specimens 
might share the derived character of unusually long, and posteriorly 
tapering, trunk centra. The shape of the centrum in the type can be 
inferred from the cross-sectional views of the vertebrae, which in 
some parts of the trunk show a similar outline to the ventral views in 
HUJ-PAL EJ699, being wide across the transverse processes and 
narrowing steeply behind them to be almost parallel-sided posteriorly. 
In MCSNT 9962, where only the upper part of the neural arch and 
postzygapophyses are visible, the vertebrae are about 3/4 as long 
(between successive neural arches) as wide (across 
postzygapophyses), which is similar to proportions in the more 
posterior part of HUJ-PAL EJ699. 

All three specimens share a distinctive feature of the ribs in that 
the distal portion, representing most of their length, is nearly straight. 
This is interpreted as a derived condition corresponding to lateral 
compression of the trunk region, as in the pachyophiids and some 
other groups of thoroughly aquatic snakes. All three specimens also 
exhibit, as far as can be seen, complete but small girdles and limbs. 
The development of the forelimb and shoulder girdle in the current 
specimen matches the development of the pelvis and hindlimb in the 
type and MCSNT specimens of Mesoleptos. The shoulder girdle and 
forelimb in HUJ-PAL EJ699 are relatively small, but complete in 
that all major elements are present. All ossified shoulder girdle 
elements except the interclavicle are preserved, while (based on the 
size and ossification of the humerus) most of the distal forelimb 
bones were present. This is consistent with the small but well 
developed (though incompletely preserved) sacrum, pelvis and 
hindlimb in the two previously known specimens of Mesoleptos. 
The observation that the shoulder girdle and forelimb in HUJ-PAL 
EJ699 are both reduced in size but complete, as is the pelvis and 
hindlimb in Mesoleptos, further suggests they are the same or 
closely related species. 

Thus, HUJ-PAL EJ699 can be associated with the two known 
specimens of Mesoleptos because ( 1 ) they exhibit no significant 
differences from each other, though they all differ from all other 
squamates, (2) they have derived similarities in the ribs (otherwise 
found only in very different forms) and, less certainly, in the 
vertebrae and limbs. 

'Mesoleptos' cf.zendrinii 

Gorjanovic-Kramberger ( 1 892: pi. Ill, fig. 4) reported a specimen in 
I. Novak's collection showing several articulated vertebrae with 
ribs, and fragments of some other elements, which he referred to 
Mesoleptos, close to M. zendrinii. The collection consisted of 
material from Cretaceous deposits of Isola di Lesina (Italian name 
for Hvar Island), Croatia (Gorjanovic-Kramberger, 1 892). This was 
held after his death by his widow Antonia Novak (Kornhuber, 1 90 1 : 
19) but the present location of this material is unknown (Calligaris, 
1988). Gorjanovic-Kramberger interpreted the specimen as exposed 
ventrally, but the shape of contacts between condyles and cotyles 
visible in his figure suggest that the vertebra may actually be 
exposed in dorsal view but sectioned horizontally at the base of the 
neural canal; this would invalidate comparisons based on the sup- 
posed ventral surface, though not the overall outline, of the centrum. 
The shape of the centrum in the most complete vertebra is very 
similar to vertebrae 9-13 of HUJ-PAL EJ699. The elongate and 






MESOLEPTOS AND THE ORIGIN OF SNAKES 



135 



posteriorly narrow centra have been regarded as diagnostic of 
Mesoleptos, and are not found in any other limbed squamates, 
though a similarly shaped centrum is present in some primitive 
snakes (e.g. Lapparentophis, Hoffstetter. 1960; Patagoniophis, 
Scanlon, 1993; Coniophis, Gardner and Cifelli, 1999). 

Girdle and limb elements are also present in the Novak specimen; 
Gorjanovic-Kramberger (1892: 99) describes 'indistinct impres- 
sions' of the humerus, radius, ulna and two metacarpals, altogether 
measuring 93.3 mm in length. This must be less than the total length 
of the forelimb, because the elements are incompletely represented 
(the ends of the long bones are obscured and the humerus can not be 
compared in detail with HUJ-PAL EJ699), but it can be concluded 
that a forelimb was present and equivalent in length to between three 
and four thoracic vertebrae, just as in the HUJ specimen. Plate-like 
structures are also shown just anterior to the supposed humerus in 
Gorjanovic-Kramberger's figure, suggesting the posterior margins 
of a scapula and coracoid like those of the HUJ specimen, although 
no useful details can be compared. 

On the other hand the ribs, although long, are curved throughout 
their length. While the centrum length of the one well-preserved 
vertebra is about 3 1 .5 mm, the length of the most complete rib 
(belonging to the preceding vertebra) is over 90 mm (Gorjanovic- 
Kramberger, 1892). These proportions seem to indicate a position 
deep within the dorsal region. In HUJ-PAL EJ699, curved ribs only 
occur up to the anterior dorsal region while more posterior ribs are 
straight. Thus the Novak specimen apparently lacks this apomorphy 
shared by the type of Mesoleptos zendrinii with the MCSNT and 
HUJ specimens (neither Gorjanovic-Kramberger nor subsequent 
writers have commented on this difference). It should therefore not 
be referred to Mesoleptos, but might possibly represent a species 
closely related to either Mesoleptos or the Mesoleptos-smxke clade 
(see below). 

Adriosaurus, Acteosaurus 

Adhosaunis suessi Seeley, 1881 (Lee and Caldwell, 2000) and 
Acteosaurus tommasinii von Meyer, 1860 (considered identical by 
Nopcsa, 1 923) are small marine lizards with distally straight ribs and 
thus, laterally compressed bodies. Adriosaurus is known from two 
specimens, from Upper Cretaceous deposits of Comen. Slovenia 
and Lesina (=Hvar), Croatia, while Acteosaurus is known from a 
single specimen from Comen. However, they both differ from 
Mesoleptos in lacking the distinctly small cervicals (relative to 
dorsals), in possessing proportionally larger limbs, proportionally 
shorter and wider dorsal vertebrae, and in exhibiting heavy 
pachyostosis of both dorsal vertebrae and ribs. They are also much 
smaller than Mesoleptos. 

Eidolosaurus trauthii 

Nopcsa (1923) described Eidolosaurus trauthii from a near-com- 
plete skeletal impression found during the demolition of a house in 
the Istrian region, i.e. in the same general region as Comen, but 
possibly within the present borders of either Slovenia, Croatia, or 
Italy (more precise locality details were not provided). This speci- 
men is currently housed in the Geologische Staatsanstalt, Vienna but 
has yet to be completely prepared. Fragments of the skull are present 
in articulation with the vertebral column, so that the total number of 
presacral vertebrae can be determined as 34. Short, slender ribs were 
present on at least three posterior cervicals, but on the basis of a 
sharp increase in length and thickness between adjacent ribs (as 
there is no trace of the sternum), Nopcsa counted 1 1 cervical and 23 
dorsal vertebrae. Two sacral and 48 or more postsacral vertebrae 
were also present. Nopcsa interpreted the type of Mesoleptos zendrinii 



as also having 23 dorsal vertebrae. The numbers of cervical and 
trunk vertebrae in Mesoleptos and Eidolosaurus are therefore com- 
parable. The relative femur length is similar in both, corresponding 
to the length of three middle dorsal vertebrae. However, there are 
also considerable differences: in Eidolosaurus the centra of trunk 
vertebrae are as wide as long, with no indication of a posterior taper; 
there is a median groove between paired ridges on the ventral 
surface throughout the trunk (the groove further divided by a median 
ridge in posterior vertebrae); all trunk ribs are strongly and uni- 
formly curved and greatly thickened; and both the vertebrae and ribs 
are pachyostotic. 

Nopcsa (1923: 107, footnote) also mentions 'An undescribed 
fossil discovered by Professor Jakel, which came to my attention 
while this work was in press, shows 1 8 posteriorly tapering vertebral 
centra, which bear long, slightly curved, proximally club-shaped 
ribs. The specimen is 28 cm long. The vertebral centra show a 
shallow but well developed median longitudinal groove. The ante- 
rior centra are almost triangular and wider than long. The general 
habitus is Mesoleptos-Yike. but the ribs are somewhat pachyostotic. 
Probably this form is related to Eidolosaurus.' 1 This may have been 
the specimen collected by Prof. O. Jakel at Lesina which Kornhuber 
(1901: 3) mentioned and referred to Carsosaurus. It would be 
particularly interesting to compare this specimen with HUJ-PAL 
EJ699, which also resembles Mesoleptos but has somewhat thick- 
ened ribs, but no illustration was provided and again the present 
location of the specimen is unknown (Calligaris, 1988: 117). 

Dolichosaurs: Dolichosaurus, Coniosaurus, 
Pontosaurus 

Dolichosaurus longicollis Owen, 1850a from the English Chalk 
(Owen, 1842, 1851; Caldwell, 2000) and Pontosaurus lesinensis 
(Kornhuber. 1873) from Hvar. Croatia, are elongate, Cenomanian 
marine squamates known from two or more articulated partial 
skeletons, and are thus important for comparison with Mesoleptos. 
They both clearly differ from HUJ-PAL EJ699 and the other 
Mesoleptos specimens in the shape of the ribs (distally curved rather 
than straight), the more gradual changes in rib length and vertebral 
dimensions along the trunk, and greater number of dorsal vertebrae, 
all of which correspond to a more slender and cylindrical body form. 
Individual mid-trunk vertebrae of Dolichosaurus differ from those 
of Mesoleptos in being less massive, and having proportionally 
larger condyles and cotyles. Otherwise, they are similar in possess- 
ing broad transverse processes, a posteriorly cylindrical centrum, 
well-developed zygosphenes, a long neural spine, and absence of 
pachyostosis. Vertebral morphology of Pontosaurus can not yet be 
adequately compared because the specimens remain incompletely 
prepared (Calligaris, 1988). Coniosaurus crassidens Owen, 1850a 
(Coniosaurus Caldwell and Cooper, 1999, invalid emendation or 
sustained lapse) and Coniosaurus gracilodens Caldwell, 1999, oc- 
cur in the same deposits as D. longicollis but comparisons are more 
problematic. Only very incomplete postcranial remains of 
Coniosaurus are known; the vertebrae are very similar to those of 
Dolichosaurus, and the two species are diagnosed by features of the 
jaws and teeth unknown in Dolichosaurus. Thus, one of the species 
of Coniosaurus might be synonymous with Dolichosaurus (Caldwell, 
2000). 

Pachyvaranus crassispondylus 

Pachyvaranus was described from the Maastrichtian of Morocco 
(Arambourg and Signeux, 1952: 288-91, pi. 41) based on a small 
number of isolated vertebrae (MNHN PMC l^t) and two doubtfully 
associated osteoderms (PMC 5-6), and originally referred to 



136 



M.S.Y. LEE AND J.D. SCANLON 



Aigialosauridae. However, it has narrower condyles and cotyles, 
relatively longer centra and more prominent transverse processes 
than known aigialosaurs. This suggests it should be compared with 
HUJ-PAL EJ699, which it resembles in size. The Pachyvaranus 
specimens are from marine phosphate deposits, and differ from 
HUJ-PAL EJ699 in the thick and compact ossification of the verte- 
brae (pachyostosis). The vertebrae also differ in that the centrum of 
Pachyvaranus is triangular, tapering rather than nearly parallel- 
sided posteriorly, but this could be a result of pachyostosis; in other 
pachyostotic reptiles the centra are further expanded posterolaterally, 
and nearly rectangular. Further, the reported 'zygosphene' is only a 
small triangular projection comparable to that of Varamis, which 
does not bear facets for articulation with a zygantrum on the 
preceding vertebra. Arambourg and Signeux considered possible 
affinities with dolichosaurs (ruled out by the lack of true zygosphenal 
articulations in Pachyvaranus) as well as aigialosaurs (noting differ- 
ences including the narrower condyles). The lack of zygosphenes in 
Pachyvaranus also rules out affinities with aigialosaurs, since recent 
studies (Carroll and DeBraga, 1992) have demonstrated the pres- 
ence of well-developed zygosphenes in aigialosaurs. However, 
affinities with Mesoleptos were not considered. No material other 
than trunk vertebrae (and doubtfully associated osteoderms) has 
been described for Pachyvaranus, and conversely the vertebrae of 
Mesoleptos are not fully known 'in the round', so that it is not yet 
possible to make detailed comparisons. 

Pachyophiidae 

Three long-bodied, limb reduced Cretaceous marine squamates 
have been referred to Pachyophiidae: Pachyophis woodwardi 
Nopcsa, 1923 (Lee et al., 1999), Mesophis nopcsai Bolkay, 1925, 
and Pachyrhachis problematicus Haas, 1979 (Haas, 1980; Lee and 
Caldwell, 1998; Zaher and Rieppel, 1999). Haas (1979) originally 
included Pachyrhachis in Simoliophiidae, as did McDowell (1987) 
who also added Pachyophis; but of the two family-group names 
proposed by Nopcsa ( 1 923), Pachyophiidae has page priority. There 
is now agreement that pachyophiids are snakes but their exact 
position within snakes remains debated (Zaher and Rieppel, 1999; 
Tchernov et al. 2000; Lee and Scanlon, 2002). These three taxa are 
extremely similar, and possess small heads, heavily pachyostotic 
mid-body vertebrae and ribs, and distally straight ribs indicating 
lateral compression of the trunk. Radovanovic (1935: 411) postu- 
lated that Mesophis was a terrestrial snake in which the very slender 
distal parts of the ribs had been straightened by pressure during 
fossilization. However, this hypothesis is very unlikely because ribs 
of similar shape occur consistently in otherwise undistorted speci- 
mens of larger pachyophiids, namely Pachyophis and Pachyrhachis, 
as well as in other marine taxa (see below). 

The specimen described here is clearly not a pachyophiid because 
in all known pachyophiids the forelimbs and shoulder girdle are 
completely absent, and the mid-trunk ribs are heavily swollen 
(pachyostotic). Also, the centra are long and taper posteriorly, unlike 
the pachyophiid condition of short centra that are of constant width 
throughout. The transverse processes also extend much further 
laterally than they do in pachyophiids. 

Haasiophis 

A new limbed Cretaceous marine snake, Haasiophis, has been 
described and interpreted to have affinities with Pachyrhachis 
(Tchernov et al., 2000) and by implication with pachyophiids as a 
group. However, certain cranial elements were apparently 
misidentified, and a reassessment of the morphology suggests that 
these taxa are not closely related, but are successive outgroups to 



crown-clade snakes (Lee and Scanlon 2002). The postcranial ele- 
ments of Haasiophis have yet to be properly described, making 
comparisions with Mesoleptos difficult. However, Haasiophis dif- 
fers from HUJ-PAL EJ699 in possessing heavy pachyostosis of the 
vertebrae, many more trunk vertebrae, and in completely lacking a 
shoulder girdle and forelimb. 

Palaeogene Marine Snakes 

HUJ-PAL EJ699 can be confidently excluded from the following 
groups of Tertiary snakes with distally straight ribs based on pres- 
ence of forelimbs and very different vertebrae. Archaeophis 
(Archaeophiinae) has long, proximally curved but distally straight 
ribs (Janensch, 1906), and the ribs of Palaeophis share this morphol- 
ogy (Owen, 1850b). However, the neural arch is narrow and high, 
the centrum approximately cylindrical and the transverse processes 
relatively small (Rage, 1984). In the complete skeleton of 
Archaeophis proavus there are over 450 trunk vertebrae and no 
traces of limbs or girdles (Janensch, 1 906), and there is no indication 
of their presence in other less completely known species. 
Anomalophis (Anomalophiidae) has similar ribs (Janensch, 1906; 
Auffenberg, 1959) and also small transverse processes. However, 
the centra are long and gradually tapering, and the neural arches are 
narrow and depressed, except for a backsloping neural spine. Verte- 
brae of other early aquatic snakes (Nigerophiidae and 
Russellophiidae; Rage, 1984, Averianov, 1997) have features re- 
sembling the palaeophiids, acrochordids and colubroids to a varying 
extent, but no ribs or articulated skeletons are known and their 
relationships remain obscure. 

Thus, the specimen HUJ-PAL EJ699 can be associated most 
closely with Mesoleptos. However, it differs from the type of M. 
zendrinii (as described by Cornalia and Chiozza, 1852; compare 
Tables 1 and 2) in the ribs of the anterior thoracic region being 
considerably shorter relative to vertebral length or width: the ribs are 
also thick in the curved middle portion of the shaft rather than 
uniformly slender. This region of the body is not preserved in the 
other referred (MCSNT) specimen. If confirmed, these differences 
would indicate a considerable variation in body shape (analogous to 
the differences among known specimens of aigialosaurids) which 
might justify erection of a new species. However, the location and 
condition of the type and some other important specimens are 
currently unknown, and the putative differences cannot be directly 
confirmed. There remains a possibility that the description and 
figure of the type are inaccurate, as they seem questionable in a 
number of details, and that the two specimens are identical. Thus, we 
have refrained from any formal taxonomic decisions pending a more 
comprehensive search for the type, and simply refer the current 
specimen to Mesoleptos sp. indet. 



RECONSTRUCTION AND PALAEOECOLOGY 

Based on all three specimens, Mesoleptos can be reconstructed as 
follows (Figs. 1 and 2). Depending on where one draws the cervical- 
dorsal boundary, there are five to seven cervical vertebrae preserved 
in HUJ-PAL EJ699, and as these do not include the atlas or axis there 
must have been at least seven to nine cervicals, and possibly several 
more. Seven to nine cervicals are plesiomorphic for squamates and 
occur in most terrestrial varanoids, aigialosaurs and some mosasaurs, 
while dolichosaurs, Eidolosaurus and some mosasaurs have increased 
from this number (Nopcsa, 1908, 1923; Caldwell. 2000). There are 
23 trunk vertebrae in the type and thus at least 30 to 32 presacrals 
altogether (cf. 34 in Eidolosaurus), but not many more than this 



MESOLEPTOS AND THE ORIGIN OF SNAKES 



137 





Fig. 2 Reconstruction of Mesoleptos in dorsal and lateral views. The head and tail are not known in any specimen and are thus conjectural. Note the 
long neck, long laterally compressed body, and short webbed limbs. Scale bar = 10cm. 



unless the neck was unusually long. Short, curved ribs are present on 
most of the cervical vertebrae, implying a narrow cylindrical neck, 
which is similar to conditions in Eidolosaanis and some dolichosaurs, 
rather than aigialosaurs which have longer ribs on most of the 
cervicals. The cervical-thoracic boundary presumably lies around, 
or immediately posterior to, the sharp increase in rib length. 

The anterior thoracic ribs are straight distally, implying a lateral 
flattening of the trunk region. Allowing for apparent variation 
between the three known specimens in the proportional length of the 
ribs, the ribs remain long throughout the mid-trunk region, where the 
largest vertebrae occur. Vertebral and rib dimensions increase stead- 
ily up to at least the tenth thoracic vertebra, are highest in mid-trunk 
and decrease, apparently more slowly, in the last ten or so presacrals. 
These size gradients are stronger than seen in measured skeletons of 
Varanus and Heloderma (Scanlon, unpublished data), and far more 
conspicuous than in any other marine varanoids described. Unlike 
some aigialosaurs and all mosasaurs, there is not a long series of 
shortened posterior dorsal ribs. Rather, long, distally straight ribs 
continue at least to within the last five presacral vertebrae (as 
indicated by the MCSNT specimen; the most posterior ribs have 
been damaged or lost in both this and the type). 

The cervical vertebrae bear prominent ventral keels and hypa- 
pophyses, which are reduced on the first two thoracics and then 
disappear. The centra of the following thoracic vertebrae are smooth 
ventrally, but posterior trunk vertebrae apparently have laterally 
paired keels defining a median trough, a feature that also occurs in 
dolichosaurs. Eidolosaurus and some aigialosaurs (but often com- 
mencing more anteriorly in the trunk). The sacral vertebrae (in the 
MCSNT specimen at least) are shorter than the immediately preced- 
ing trunk vertebrae, and are fused (or at least very tightly articulated) 
together. Parts of the first few caudal vertebrae are present in the 
type, indicating a laterally compressed tail with elongate but antero- 
posteriorly narrow neural spines and chevrons. 

The trends in vertebral size (length and width) and rib length 
indicate an animal with a relatively small head and narrow neck in 
relation to its body, similar to dolichosaurs and Eidolosaurus. The 
curved cervical ribs indicate that the cervical region of the animal 
was approximately round in cross-section. However, the distally 
straight dorsal ribs indicate that the trunk region of the animal was 
laterally compressed and very deep. These long ribs projected only 
a short distance laterally from the vertebrae before curving to extend 
downward (and obliquely backward) for most of their length. The 
girdles and limbs were rather small, although most elements were 
probably present; compared to adjacent vertebrae, both the femur 



and humerus are relatively shorter than in aigialosaurs, but the 
forelimb was not as reduced as in dolichosaurs or Eidolosaurus (Fig. 
2). 

In comparison with the similar-sized aigialosaurs, Pontosaurus 
and species of Varanus. trends in vertebral size within the column of 
Mesoleptos are somewhat different. There is a local minimum of 
centrum length in the posterior cervical region, but the elongation of 
the anterior cervicals is much less pronounced than the condition in 
most Varanus spp. (a derived condition within that genus). Gradi- 
ents of vertebral length and width within the thoracic and dorsal 
region are stronger than in any of the other taxa. The centrum is 
narrower posteriorly than in aigialosaurs, Pontosaurus and Varanus, 
indicating a condyle-cotyle joint of smaller diameter and surface 
area. This in turn suggests weaker compressive forces within the 
column, along with a less energetic style of locomotion and/or a 
greater capacity for lateral flexion of the neck and trunk. On the 
other hand, the combination of long transverse processes and long 
narrow centra increases both leverage and space for muscles con- 
necting successive transverse processes, such as the m. 
interarticularis (cf. Mosauer, 1935; Gasc, 1974). These could then 
be of increased importance in lateral undulation, perhaps taking over 
in this role from longer muscles inserting on the ribs whose effec- 
tiveness would be decreased by lateral compression of the trunk. If 
the above interpretation of the affinities of the Novak specimen is 
correct, the derived vertebral morphology evolved before the lateral 
compression, so that this 'takeover' could happen via an intermedi- 
ate where both sets of muscles were effective. Zygosphenes, 
considered to be of biomechanical importance in limiting twisting 
between adjacent vertebrae (Gasc, 1974), are well-developed (ex- 
posed dorsally in the MCNST specimen) and articulate with zygantra 
in the preceding neural arches as in other aquatic varanoids and all 
snakes. 

Among living squamates, the only forms with distally straight 
ribs (and thus laterally compressed bodies) are highly aquatic 
caenophidian snakes, such as file snakes (acrochordids) and sea 
snakes (laticaudine and hydrophiine elapids). This feature has rarely 
been discussed in extant snakes; Hoffstetter and Gayrard (1965) do 
not comment on any unusual features of the ribs in Acrochordus or 
Enhydrina (Hydrophiinae), though it was described in 'Enhydris' 
(=Lapemis) hardwickii (Hydrophiinae) by Janensch (1906: 22). In 
Acrochordus arafurae (SAM R26956, R26966) the anterior ribs are 
robust and strongly curved, while those of the posterior half of the 
body are much more slender and only weakly curved except near the 
base. The pachyophiids - primitive marine snakes - also had a 



138 



M.S.Y. LEE AND J.D. SCANLON 



similar morphology, which is functionally correlated with 
anguilliform swimming (Scanlon et ah, 1999; Lee et al., 1999). It 
can thus be concluded that Mesoleptos was marine. This is also 
supported by the morphology of the posterior ribs . While they do not 
exhibit the histological features of true pachyostosis, they are never- 
theless robust (in the MCSNT and HUJ specimens) and might have 
served to reduce bouyancy, much like the pachyostotic ribs in other 
marine reptiles. The type of M. zendrinii, from the Comen locality, 
also comes from deposits dominated by marine fish (Gorjanovic- 
Kramberger, 1892) and is associated with aigialosaurs, dolichosaurs 
and pachyophiids. This also presumably applies to the MCSNT 
specimen (although its collection details have not been recorded it is 
probably from either Comen or Lesina). Marine habits of the present 
specimen are also implied by the position of the 'Ein Jabrud locality 
far from the palaeoshoreline (Scanlon et al., 1999), and the articu- 
lated nature of the preserved elements suggesting in situ preservation. 
The laterally compressed body and small limbs suggest that 
Mesoleptos swam primarily by lateral undulation, holding its limbs 
against its flanks (Carroll 1985; Lee 1999). In such forms, most of 
the propulsion occurs by movements of the tail, and to some extent 
the posterior region of the trunk. This is consistent with the observa- 
tion that the posterior trunk region is most laterally compressed in 
Mesoleptos (the tail is unknown). The forelimbs and hindlimbs, 
however, were still large and well ossified enough to have been 
functional. They may have been used for slow locomotion ('walk- 
ing') along the seabed, where (with the help of buoyancy) they could 
have supported the body. Alternatively, or additionally, they may 
have been used for forays on the shore. 



PHYLOGENETIC RELATIONSHIPS OF 
MESOLEPTOS 

All previous interpretations of the morphology and relationships of 
Mesoleptos were based either on poorly preserved and inadequately 
described material (the type), or on a composite of the type with the 
referred Novak specimen (Gorjanovic-Kramberger, 1892) which is 
clearly distinct from M. zendrinii in rib morphology. Gorjanovic- 
Kramberger's inclusion of Mesoleptos in Varanidae was 
'phenetically' based on its long ribs, as distinct from the shorter and 
more uniform ribs of Aigialosaums (as then interpreted) and 
dolichosaurs. However, he recognised it as marine in habits and thus 
by no means a typical varanid. Nopcsa classified Mesoleptos doubt- 
fully as an aigialosaur (1903), but later placed it in a separate 
subfamily (Mesoleptinae) with Eidolosaurus, close to both 
aigialosaurines and dolichosaurines within Dolichosauridae (1923). 

McDowell and Bogert (1954) returned Mesoleptos and 
Eidolosaurus, again doubtfully, to Aigialosauridae, but also briefly 
considered that they might be related to the living earless monitor, 
Lanthanotus. Hoffstetter (1955) also retained Mesoleptos as a poss- 
ible aigialosaurid, while recognising Eidolosaurus as a dolichosaur 
and suggesting that Pachyvaranus might represent a distinct family. 
Romer (1956) placed both Mesoleptos and Eidolosaurus, with 
question marks, in Dolichosauridae. 

The current state of understanding of these groups is perhaps best 
indicated by the fact that the systematic conclusion to Calligaris' 
(1988) review was formed by a summary of Nopcsa's (1923) 
classification, without substantial additions or revisions. That these 
groups have been poorly studied recently is highlighted by Carroll 
and DeBraga's (1992) statement that only five species had been 
assigned to Aigialosauridae, and did not mention either Mesoleptos 
(assigned to Aigialosauridae by Nopcsa, 1903, Camp, 1923, and 



McDowell and Bogert, 1954), Eidolosaurus (assigned by Nopcsa, 
1923, and McDowell and Bogert, 1954) or Pachyvaranus (assigned 
by Arambourg and Signeux, 1952). 

The relationships of Mesoleptos, therefore, remain unresolved. 
While a robust assessment will have to await more complete material, 
a preliminary analysis is undertaken here. Morphological informa- 
tion from the MCSNT and HUJ specimens (based on examination of 
specimens) and the type (based on published descriptions) was used 
in order to evaluate its phylogenetic relationships. Mesoleptos was 
added to the data matrix used in the most recent comprehensive 
analysis of squamates (Lee 2002); this matrix includes 248 osteo- 
logical characters, used here, and addresses recent criticisms of 
various characters (Rieppel and Zaher 2000). Recently described (or 
redescribed) elongate marine squamates were also included in this 
matrix: Pachyrhachis (Lee and Caldwell, 1998), Pachyophis (Lee et 
al., 1999), Adriosaurus (Lee and Caldwell, 2000) and dolichosaurs 
(Coniosaurus and Dolichosaurus; Caldwell and Cooper, 1999; 
Caldwell, 1999, 2000). Coniosaurus and Dolichosaurus are here 
combined into a single taxon, Dolichosauridae sensu stricto, based 
on the observations that the comparable parts of the two taxa appear 
almost identical, they overlap stratigraphically, and as noted by 
Caldwell (2000) one of the Coniosaurus species might be synony- 
mous with D. longicollis. Character codings for all taxa (except 
Haasiophis) in this matrix, including the marine fossil forms, are 
based on direct examination of the material. As descriptions of the 
remaining marine squamates discussed above are dated, and they 
have yet to be restudied, they have not been included in the analysis. 
The full matrix is presented elsewhere (Lee 2002) and only the 
(new) character codings for Mesoleptos are listed here (Appendix 
1 ). The full matrix (including Mesoleptos) used in this analysis has 
been deposited in TreeBase (http://www.treebase.org/treebase/). 

The enlarged data matrix with Mesoleptos was analysed using the 
heuristic algorithm of PAUP* (Swofford, 1999) employing 100 
random addition sequences. Two analyses were performed, with 
multistate characters ordered according to morphoclines where 
possible, or with all multistate characters unordered, to see if the 
phylogenetic analyses were contingent on assumptions of character 
state transitions. The degree of support for each grouping was 
ascertained by the support index (Bremer, 1988), calculated in 
PAUP using batch commands generated by TreeRot Version 2b 
(Sorenson, 2000). These commands were modified so that each 
heuristic search employed 100 rather than 20 random addition 
sequences. Nonparametric bootstrapping (1000 heuristic replicates 
each employing 100 random addition sequences) was also used to 
assess the robustness of each clade. As there were no fully specified 
a priori hypotheses for Mesoleptos and all other squamates, 
Templeton tests are inappropriate and were not performed (Goldman 
era/., 2000). 

Phylogenetic affinities 

In the ordered analysis, three most parsimonious trees were found, 
each of length 672, consistency index = 0.46, retention index = 0.7 1 . 
The strict consensus is shown in Fig. 3A, along with nodal supports. 
In the unordered analysis, 4 most parsimonious trees were found 
when only branches with unequivocal character support were re- 
tained, each of length 639, consistency index = 0.48, retention index 
= 0.71. The strict consensus is shown in Fig. 3B. along with nodal 
supports. 

The basic topologies of the ordered and unordered consensus 
trees are similar to each other and largely unchanged from that the 
previous study (Lee, 2002), so that diagnoses of all the clades within 
Squamata are not repeated here. The characters diagnosing additional 



MESOLEPTOS AND THE ORIGIN OF SNAKES 



139 




Clade D: 2,82 
Ophidia: 5,98 
Clade C: 1,66 
Clade B: 1,50 

Clade A: 3,70 

Pythonomorpha: 7,99 
Thecoglossa: 9,92 
Varanoidea: 7,95 
2,61 
Anguimorpha: 4,75 



Scleroglossa: 6,93 



Ophidia: 4,96 
B: 1,46 




Clade A: 5,79 

nomorpha: 7,99 
Thecoglossa: 7,85 
Varanoidea: 7,91 
1,51 
uimorpha: 1,57 



Fig. 3. The phylogenetic affinities of Mesoleptos, based on cladistic analyses of 248 characters across squamates. (A) Analysis with multistate characters 
ordered, strict consensus of 3 trees, length 672, consistency index 0.46, retention index 0.71. (B) Analysis with multistate characters unordered, strict 
consensus of 3 trees, length 639, consistency index 0.48, retention index 0.71. First number next to each clade refers to branch support (Bremer, 1988); 
second number refers to bootstrapping frequency. Clades immediately relevant to the affinities of Mesoleptos are in bold and are discussed in the text. 
The other more inclusive clades are discussed in Lee (1998). Aquatic terminal taxa are indicated in bold. 



140 



M.S.Y. LEE AND J.D. SCANLON 



clades of immediate relevance to Mesoleptos are listed below. The 
character changes diagnosing these clades in the 'ordered' analysis 
under delayed transformation optimisation are listed in Appendix 2; 
the changes in the 'unordered' analysis are very similar, except that 
some clades collapse (compare Figs 3 A and B ). Unequivocal changes, 
i.e. those which occur under both delayed and accelerated 
optimisation, are indicated with an asterisk (*). Note that most 
characters diagnosing snakes (Ophidia) are equivocal because 
Mesoleptos, the sister group of snakes, is poorly known and nearly 
all the characters could apply to a more inclusive clade that also 
contains Mesoleptos (clade C). As discussed below, the clades are 
not strongly corroborated due to missing data and possible correla- 
tion of the supporting characters, and are thus not yet named 
formally. 



EVOLUTIONARY IMPLICATIONS 

The phylogenetic results imply that snakes arose from within a 
plexus of marine varanoids, an idea suggested initially by Nopcsa 
(1908, 1923) and later by Haas (1980). The aquatic hypothesis is 
often ascribed to Cope (1869), but Cope never suggested that the 
aquatic mosasaurs were ancestral to snakes: rather, he suggested that 
both had a close common ancestor, which might even have been 
terrestrial. However, critics subsequently misquoted Cope as sug- 
gesting that snakes evolved directly from mosasaurs and thus had 
marine ancestors, and then proceeded to argue that as snakes could 
not have evolved from mosasaurs (which possess numerous 
specialisations), they could not have had marine ancestors (e.g. 
Owen, 1877; Dollo, 1903, 1904; Janensch, 1906; McDowell and 
Bogert, 1954; Zaher and Rieppel, 1999). Nopcsa (1908, 1923) 
recognised and addressed the erroneous arguments of Owen and 
Janensch, and put forward a rigorous case for a marine stage in snake 
ancestry. More recently, by interpreting aigialosaurs as probable 
ancestors of snakes, McDowell and Bogert (1954) implicitly pro- 
posed a marine ancestry. In describing the second specimen of 
Pachyrhachis (=Ophiomorphus), Haas (1980: 191) stated that the 
fossil 'points to the fact that the snakelike body and loss of limbs did 
develop in a marine surrounding'. Despite this, the aquatic theory 
has in recent times been largely rejected in favour of the 'fossorial 
theory', i.e. that snakes evolved from small elongate burrowing 
lizards (e.g. Janensch, 1906; Walls, 1940; Bellairs and Underwood, 
1951; Underwood, 1967; Rieppel, 1988; Greene, 1997). Thus, few 
modern studies rigorously surveyed marine varanoids and marine 
ophiomorphs for possible relationships with modern snakes. 

This analysis indicates that the closest four to eight outgroups to 
modem (terrestrial) snakes are marine; the exact number varies 
depending on how the polytomies are resolved. The most parsimo- 
nious interpretation is that marine or at least semi-aquatic habits 
were primitive for pythonomorphs, and that snakes evolved in a 
marine or semi-aquatic environment and are secondarily terrestrial 
(Nopcsa, 1908, 1923; McDowell and Bogert, 1954; Haas, 1980). 
In order to maintain that the snake stem lineage was always terres- 
trial, between four and eight convergent invasions of marine 
habitats must be assumed to have occurred in mosasauroids, 
dolichosaur-like taxa, and basal snakes. The analysis further sug- 
gests that, of all the marine varanoids, Mesoleptos occupies a 
crucial phylogenetic position, as the nearest relative of snakes 
(Ophidia). If this is true, the similarities between Mesoleptos and 
primitive snakes are not convergent; these include such traits as a 
proportionally small head, long body, limb reduction, and lateral 
body compression. In these features, Mesoleptos appears inter- 



mediate between the typical lizard-like marine varanoids (e.g. mosa- 
saurs) and primitive marine snakes. 

Two substantial caveats must be added to this interpretation. 
Apart from mosasaurs and aigialosaurs, all the marine varanoids are 
very imperfectly known. For instance, Mesoleptos can be scored for 
only 13% of characters, dolichosaurs for 35% and Adriosaurus for 
38%. Such large amounts of missing information suggest that their 
positions cannot be very robust, a view confirmed by low bootstrap 
and Bremer supports. This missing information also reduces support 
throughout the tree, as the poorly known taxa can fit into many 
different places with only slight loss in parsimony. Additionally, 
many of the characters that unite dolichosaurs, Adriosaurus and 
Mesoleptos with mosasauroids and snakes, to the exclusion of other 
varanoids, are correlates of marine adaptation. Within this group 
(Pythonomorpha), many of the characters uniting dolichosaurs, 
Aphanizocnemus, Adriosaurus and Mesoleptos with snakes to the 
exclusion of mosasauroids are correlates of body elongation and 
limb reduction. Thus, the position of these poorly known taxa close 
to snakes might reflect a false signal caused by marine adaptation 
and body elongation, both features found in basal snakes. More 
complete fossil finds, and thus, information on characters not obvi- 
ously correlated with habitat and body form, are required before 
their phylogenetic relationships can be conclusively ascertained and 
the early evolution of snakes clearly understood. The fundamental 
questions investigated by Bellairs and Underwood (1951) and 
Underwood (1967) regarding the affinities and ecological origins of 
snakes still await convincing answers. 



ACKNOWLEDGEMENTS. ML thanks Garth Underwood for assistance, 
friendship, and inspiration during his visits to the Natural History Museum. 
We also thank Harry Greene. David Cundall. Michael Caldwell, Jenny Clack 
and Susan Evans for comments and/or discussion, and Eitan Tchernov 
(Hebrew University, Jerusalem) for hospitality in Israel and allowing study 
and description of their specimen of Mesoleptos. Supported by an Australian 
Research Council Senior (QEII) Fellowship and Research Grant to ML. 



REFERENCES 



Arambourg, C. & Signeux, J. 1952. Les vertebres fossiles des gisements de phos- 
phates (Maroc - Algerie - Tunisie). Sennce Geologique du Maroc, Notes et Memoires 
92: 1-359. 

Auffenberg, W. 1 959. Anomalophis bolcensis (Ma$sa\ongo). a new genus of fossil snake 
from the Italian Eocene. Breviora, Museum of Comparative Zoology 114: 1-16. 

Averianov, A.O. 1997. Paleogene sea snakes from the eastern part of Tethys. Russian 
Journal of Herpetology 4: 128-142. 

Bellairs, A. D'a. & Underwood, G. 1 95 1 . The origin of snakes. Biological Reviews of 
the Cambridge Philosophical Society 26: 193-237. 

Bolkay, S.J. 1925. Mesophis nopcsai n. g„ n. sp.; ein neues. schlangenahnliches Reptil 
aus der unteren Kreide (Neocom) von Bilek-Selista (Ost-Hercegovina). Glasnika 
Zemaljskog Museja it Bositi i Hercegovini 37: 125-136. 

Boulenger, G.A. 1891. Notes on the osteology of Heloderma horridum and H 
suspectum, with remarks on the systematic position of the Helodermatidae and on the 
vertebrae of the Lacertilia. Proceedings of the Zoological Society of London 1891: 
109-121. 

Bremer, K. 1988. The limits of amino-acid sequence data in angiosperm phylogenetic 
reconstruction. Evolution 42: 795-803. 

Caldwell, M.W. 1999. Squamate phylogeny and the relationships of snakes and 
mosasauroids. Zoological Journal of the Linnean Society 125: 1 15-147. 

. 2000. On the aquatic squamate Dolichosaurus longicollis Owen, 1850 

(Cenomanian, Upper Cretaceous), and the evolution of elongate necks in squamates. 
Journal of Vertebrate Paleontology 20: 720-735. 

, & Cooper, J. 1999. Redescription. palaeobiogeography, and palaeoecology of 

Coniasaurus crassidens Owen, 1850 (Squamata) from the English Chalk (Creta- 
ceous, Cenomanian). Zoological Journal of the Linnean Society 127: 423^t52. 

Calligaris, R. 1988. I rettili fossili degli 'Strati calcarei ittiolotici di Comen' e 
dell'Isola di Lesina. Atti del Museo Civico di Storia Naturale, Trieste 41: 85-125. 



MESOLEPTOS AND THE ORIGIN OF SNAKES 



141 



Camp, C.L. 1923. Classification of the lizards. Bulletin of the American Museum of 
Natural History 48: 289-481. 

Carroll, R.L. 1985. Evolutionary constraints in aquatic diapsid reptiles. Special 
Papers in Palaeontology 33:145-155. 

Carroll, R.L. & Debraga, M. 1992. Aigialosaurs: mid-Cretaceous Varanoid lizards. 
Journal of Vertebrate Paleontology 12: 66-86. 

Cope, E.D. 1869. On the reptilian orders Pythonomorpha and Streptosauria. Proceed- 
ings of the Hasten Society of Natural History 12: 250-266. 

Cornalia, E. & Chiozza, L. 1852. Cenni geologic] sull' Istria. Giomale dell' 1. R. 
Institute Lombardo 3: 1-35. 

Dollo, L. 1903. Les ancetres des Mosasauriens. Bulletin scicniifique de France el 
Belgique 38: 137-130. 

1904. L'origine des Mosasauriens. Bulletin du Societe belgique de Geologie, 

Hydrologie et Palaeontologie, Bruxelles 18: 217-222. 

Estes, R., De Queiroz, K. & Gauthier, J. 1988. Phylogenetic relationships within 

Squamata. In R. Estes and G. Pregill (eds). Phylogenetic relationships of the lizard 

families. Stanford University Press, Stanford, California. Pp. 119-281. 
Gardner, J.D. & Cifelli, R.L. 1999. A primitive snake from the Cretaceous of Utah. 

Special Papers in Palaeontology 60: 87- 1 00. 
Gasc, J.-P. 1974. L' interpretation fonctionelle de lappareil musculo-squelettique de 

1' axe vertebral chez les serpents (ReptWtd). Memoires du Museum nalionale d'Histoire 

natu relic. Series A 83: 1-182. 
Goldman, N, Anderson, J. P. & Rodrigo, A.G. 2000. Likelihood-based tests oi 

topologies in phylogenetics. Systematic Biology 49: 652-670. 
Gorjanovic-Kramberger, C. 1892. Aigialosaurus, eine neue Eidechse aus die 

Krcideschiefern der Insel Lesina mil Rucksicht auf die bereils beschriebenen 

Lacertiden von Comen und Lesina. Glasnik huvatskoga naravoslovnoga drustva 

(Societas historico-naturalis croatica) u Zagrebu 7: 74-106. 
Greene, H.W. 1997. Snakes - The evolution of mystery in nature. University of 

California Press. Berkeley. 
Haas, G. 1979. On a new snakelike reptile from the lower Cenomaniai) of Ein Jabrud, 

near Jerusalem. Bulletin du Museum national d'Histoire naturelle. Paris (4). Section 

CI: 51-64. 

1980. Remarks on a new ophiomorph reptile from the lower Cenomanian of Ein 

Jabrud. Israel. In L. L. Jacobs fed.). Aspects of vertebrate history: Essays in Honour 
of E. H. Colbert. Museum of Northern Arizona Press: Flagstaff AZ. Pp. 177-202. 

Hoffstetter, R. 1955. Squamates de type moderne. Trade de Palaontologie 5: 606-662. 

1960. Un serpent terrestre dans le Cretace inferieurdu Sahara. Bulletin du Sot ieti 

giologique de France 7: 897-902. 

& Gayrard, Y. 1965. Observations sur I'osteologie et la classification des 

Acrochordidae (Serpentes). Bulletin du Museum national d'Histoire naturelle. Paris 
(series 2) 36: 677-696. 

Janensch, W. 1906. Uber Archaeophis proavus Mass.. eine Schlange aus dem Eoc.m 

des Monte Bolca. Beitriige zur Palaontologie und Geologic Osterreich-Ungams 

iiml des Orients 19:1-33. 
Kluge, A.G. 1987. Cladistic relationships in the Gekkonoidea (Squamala. Saunal. 

Miscellaneous Publications of the Museum of Zoology, University of Michigan 173: 

1-54. 
Kornhuber, A. 1 873. Ubereinen neuen fossilen Saurier aus Lesina. Abhandlungen der 

k. k. geologischen Reichsanstalt 5: 75-90. 2 plates. 
Kornhuber, A. 1901. Opetiosaurus bticchichi. Eine neue fossile Eidechse aus der 

unteren Kreide von Lesina in Dalmatien. Abhandlungen der k. k. geologischen 

Reichsanstalt 17: 1-24. 
Lee, M.S.Y. 1998 Convergent evolution and character correlation in burrowing rep- 
tiles: towards a resolution of squamate phylogeny. Zoological Journal oj the 1 nine an 

Society 65: 369-453. 
1999. Aquatic reptiles. In R. Singer and M. K. Diamond (eds). Encyclopedia of 

Paleontology. Fitzroy Dearbon: London. Pp. 96-101. 
2002. Squamate phylogeny. taxon sampling, and data congruence. Hcrpelologicul 

Monographs in review. 

& Caldwell, M.W. 1998. Anatomy and relationships of Pachyrhachis 

problematicus. a primitive snake with hindlimbs. Philosophical Transactions oi the 
Royal Society of London B 353: 1521-1 552. 

& 2000. Adriosaurus and the affinities of mosasaurs. dolichosaurs and 

snakes. Journal of Paleontology 74: 915-937. 

, & Scanlon, J.D. 1999. A second primitive marine snake: Pachyophis 

woodward! from the Cretaceous of Bosnia-Herzegovina. Journal of Zoology, Lon- 
don 248: 509-520. 

& Scanlon, J.D. 2002. Snake phylogeny based on osteology, soft anatomy and 

ecology. Biological Reviews of the Cambridge Philosophical Society 77. in press. 

McDowell, S.B. 1987. Systematics. In R.A. Seigel. J.T.C. Collins and S.S. Novak 
(eds.). Snakes: Ecology and Evolutionary Biology. MacMillan. New York. Pp. 1-50. 

& Bogert, CM. 1954. The systematic position of Lanthanotus and the affinities 

of the Anguimorphan lizards. Bulletin of the American Museum of Natural History 
105: 1-154. 

Mosauer, W. 1 935. The myology of the trunk region of snakes and its significance for 
ophidian taxonomy and phylogeny. University of California Publications in Biologi- 
cal Sciences 1 : 8 1 - 1 20. 



Nopcsa, F. 1903. Uber die Varanusartigen Lacertenlstriens. Beitriige zur Palaontologie 
und Geologie Osterreich-Ungams und des Orients 15: 3 1^42. 

1908. Zur Kenntnis der fossilen Eidechsen. Beitriige zur Palaontologie und 

Geologie Osterreich-Ungams unci des Orients 21: 33-62. 

1923. Eidolosaurus und Pachyophis. Zwei neue Neocom-Reptilien. 

Palaeontographica 65: 99-154. 
Owen, R. 1842. Description of the remains of a bird, tortoise and lizard from the Chalk 

of Kent. Transactions of the Geological Society of London. Series 2. 6: 411 — 4 1 3- 
1850a. Description of the fossil reptiles of the Chalk Formation. In F. Dixon (ed.). 

The Geology and Fossils of the Tertian and Cretaceous Formations of Sussex. 

Longman. Brown. Green and Longman, London. Pp. 378^t00. 403-404. 
1850b. Monograph on the Reptilia of the London Clay. Part III. Order Ophidia. 

British Museum: London 

1851. .4 monograph on the fossil Reptilia of the Cretaceous formations. Part 1. 

The Palaeontological Society: London. 

1877. On the rank and affinities of the reptilian class of the Mosasauridae. 

Gervais. Quarterly Journal of the Geological Society oj London 33: 682-715. 

Radovanovic, M. 1935. Anatomische Studien am Schlangenkopf. Jenaische Zeitschrift 

fiir Naturwissenschaft 69: 321-422. 
Rage, J.C. 1984. Hanilbttch der Paldoherpetologie. Ted II. Serpentes. Gustav Fischer. 

Stuttgart. 
Rieppel, O. 1988. A review of the origin of snakes. Evolutionary Biology 22: 37-130. 
& Zaher, H. 2000. The intramandibular joint in squamates. and the phylogenetic 

relationships of the fossil snake Pachyrhachis problematicus Haas. Fieldiana (Geol- 
ogy) 43: 1-69. 
Romer, A.S. 1956. Osteology of the reptiles. University of Chicago Press, Chicago. 
Scanlon, J.D. 1993. Madtsoiid snakes from the Eocene Tingamarra Fauna of eastern 

Queensland. Kaupia: Oarinstddtcr Beitriige zur Naturgcschichtc 3: 3-8. 
1996. Studies in the palaeontology and systematics of Australian snakes. Unpub- 
lished Ph.D. thesis. University of New South Wales. Sydney. 
& Lee, M.S.Y. 2000. The Pleistocene serpenl Wonambi and the early evolution of 

snakes. Nature 403: 416420. 
, . Caldwell, M.W. and Shine, R. 1999. Ecology and laphonomy of the 

primitive snake Pachyrhachis. Historical Biology 13: 127-150. 
Seeley, H.G. 1 88 1 . On remains of a small li/ard from the Neocomian rocks of Comen. 

near Trieste, preserved in the Geological Musem of the University of Vienna. 

Quarterly Journal of the Geological Society of London 37: 52-56. 
Sorenson, M. 2000. TreeRot Version 2b. Computer program and documentation. 

Distributed b\ the author. Boston University. 
Swofford, D. L. 1999. PAUP*: Phylogenetic Analysis Using Parsimony (*and other 

methods). Version 4. Sinauer Associates, Sunderland MA. 
Tchernov, E., Rieppel, ()., Zaher, H., Polcyn, M. & Jacobs, L.L. 2000. A fossil snake 

with limbs. Science 287: 2010-2012 
Underwood, G. 1967. A Contribution to the Classification of Snakes. British Museum 

(Natural History) Publication No. 653. Trustees of the British Museum (Natural 

History i. I .ondon. 179pp. 
von Meyer, H.V. I860, Actaeosaurut tommasinii aus dem schwarzen Kreide-Schiefer 

von Comen am Karste. Palaeontographica 7: 223-231. 
Walls, G.L. 1940. Ophthalmological implications lor the early history of snakes. 

Copeia 1940: 1-8. 
Zaher, H. & Rieppel, O. 1999. The phylogenetic relationships of Pachyrhachis 

problematicus, and the evolution of limblessness in snakes (Lepidosauria, Squamata). 

Comptes vendues de I' Academic des Sciences de Paris. Sciences de la terre el des 

planetes 329: 831-837. 



Appendix 1 

Additions to the osteological data matrix of Lee (2002) used in this 
cladistic analysis. Mesoleptos was added to the taxon list and could 
be coded for the following characters. 

Axial Skeleton 

171. Centra. Not constricted anterior to condyle, i.e. condyle not 
wider than posterior end of centrum (0): slightly constricted 
anterior to condyle, i.e. condyle slightly wider than posterior 
end of centrum ( 1 ); greatly constricted anterior to condyle, i.e. 
condyle much wider than posterior end centrum (2). Mesoleptos 
0. 

172. Vertebral articular surfaces. Vertical, condyles (if present) 
facing posteriorly, much of the articular surface is visible in 
ventral view (0); slightly anterodorsal, condyles facing slightly 
dorsally, only the ventral edge of the articular surface is visible 



142 



M.S.Y. LEE AND J.D. SCANLON 



in ventral view ( 1 ); anterodorsal, condyles facing very dorsally, 
none of the articular surface is visible in ventral view (2). 
Mesoleptos 0. 

174. Centra. Notochordal, i.e. perforated by persistent notochord 
in adults (0); not notochordal, i.e. not perforated by persistent 
notochord in adults (1). Mesoleptos 1. 

175. Centra. Amphicoelous (0); procoelous (1). Mesoleptos 1. 

176. Neural spines. Tall processes (0); low ridges ( 1 ). Mesoleptos 0. 

177. Zygosphenes and zygantra. Present (0); absent (1). 
Mesoleptos (Visible on the MCSNT specimen). 

179. Intercentra on dorsal (thoracolumbar) vertebrae. Present 
(0); absent (1). Mesoleptos 1. 

1 80. Number of presacral vertebrae. 23 to 25 (0); 22 or fewer ( 1 ); 
26 (2); 27 to 50 (3); 50 to 119(4); 120 to 150(5), 150 or more 
(6). Mesoleptos 3. 

182. Transverse processes of cervicals. On anterior end of cen- 
trum (0); on middle of centrum (1). Mesoleptos 0. 

1 83. Hypapophyses on anterior presacrals. Only extending to the 
posterior end of the sixth presacral at most (0); extending to the 
seventh presacral or beyond (1). Mesoleptos 1. 

185. Anterior presacral vertebrae (excluding atlas and axis 
intercentra). Not sutured or fused to preceding centrum (0); 
sutured to preceding centrum (1); fused to preceding centrum 
(2). Mesoleptos 2. 

1 86. Anterior presacral vertebrae Not sutured or fused to follow- 
ing centrum (0); sutured to following centrum (1); fused to 
following centrum (2). Mesoleptos 0. 

1 87. Pachyostosis of mid-dorsal vertebrae and ribs. Absent (0); 
present (1). Mesoleptos 0. 

1 96. Body shape. Round, dorsal ribs smoothly curved (0); laterally 
compressed, middle and distal regions of dorsal ribs totally 
straight (1). Mesoleptos 1. 

198. Ribs. Proximal end without anteroventral pseudotuberculum 
(0); proximal end with anteroventral pseudotuberculum (1). 
Mesoleptos 0. 

199. Ribs. Proximal end without posterodorsal pseudotuberculum 
(0); proximal end with posterodorsal pseudotuberculum (1). 
Mesoleptos 0. 

200. Distally forked cloacal ribs ('lymphapophyses'). Absent (0); 
present (1). Mesoleptos 0. 

201. Tail. Cylindrical or only slightly lateral compressed, trans- 
verse processes well-developed, chevrons and neural spines 
not elongated (0); very laterally compressed, transverse proc- 
esses reduced anteriorly and absent posteriorly, chevrons and 
neural spines elongated (1). Mesoleptos 1. 

203. Scapulocoracoid. Present and large (0); present but reduced 
(1); absent (2). Mesoleptos 1. 

204. Emargination on anterodorsal edge of scapula. Absent (0); 
present (1). Mesoleptos 0. 

205. Anterior (primary) coracoid emargination. Absent (0); 
present (1). Mesoleptos 1. 

206. Posterior (secondary) coracoid emargination. Absent (0); 
present (1). Mesoleptos 1. 

207. Clavicle. Present (0); absent (1). Mesoleptos (see text for 
discussion of identity of this element). 

209. Clavicles. Rod-like, at most only slightly expanded proxi- 
mally and with no notch or fenestra (0); greatly expanded 
proximally, usually with notch or fenestra (1). Mesoleptos 
(see text for discussion of identity of this element). 

218. Forelimbs. Large (0); small (1), absent (2). Mesoleptos 1. 

222. Pelvis. Present and large (0); present and small ( 1 ); absent (2). 
Mesoleptos 1. 

228. Hindlimbs. Well-developed (0); reduced (1); absent (2). 



Mesoleptos 1. 

234. Body proportions. Head moderately large with respect to 
wide trunk region (0); head extremely small with respect to 
wide trunk region (1 ). Mesoleptos 1. 

235. Dorsal body osteoderms. Absent (0); present ( 1 ). Mesoleptos 
0. 

236. Ventral body osteoderms. Absent (0); present ( 1 ). Mesoleptos 
0. 

246. Epiphyses. Present on appendicular and axial skeleton (0); 
present on appendicular, but absent on axial skeleton (1); 
absent from both appendicular and axial skeleton. Mesoleptos 
1. Note Haasiophis was incorrectly coded with state 1 in Lee 
(2002); it has state 2. 

Appendix 2 Synapomorphies for Clades A-D in Fig. 3A 

Character number, consistency index and direction of change (if not 
— > 1) listed in parentheses. 

Clade A: dolichosaurs, Adriosaurus, Mesoleptos, and 
Ophidia 

*More than ten cervical vertebrae (181, 0.6, 2— »3), *Scapulocoracoid 
reduced (203, 0.67), *interclavicle absent (210, 0.33), *forelimbs 
small (218, 0.5), *pelvis reduced (222, 0.5), *hindlimbs small (228, 
0.67). 

Clade B: Adriosaurus, Mesoleptos, and Ophidia 

*Premaxilla-maxilla contact mobile (5. 1.0), frontals paired (26. 
0.14), postorbitofrontal ventral process large (36, 0.5, 1— »0), su- 
pratemporal superficial (52, 0.67, 1 — »0), *body laterally compressed 
(196, 0.5), pubis not expanded distally (227, 1.0). 

Clade C: Mesoleptos and Ophidia (snakes) 

* Vertebral condyles facing posteriorly ( 172, 1.0, 1— »0), *head small 
with respect to trunk (234, 1.0). 

Clade D: Ophidia (snakes) 

Dorsal process of maxilla on middle or anterior of maxilla (8, 0.5, 
1— >0), posterior process of maxilla long (10, 0.33, 1— >0), lacrimal 
absent (11, 0.5), frontoparietal suture with sinuous contact (30, 0.2, 
1— >0), pineal foramen absent (40. 0.17), parietal table reduced to 
sagittal crest (42, 0.22, 1 — >0), suspensorial ramus of parietal reduced 
(44, 0.5), upper temporal arch incomplete (45, 0.25), tympanic crest 
absent (57, 0.29, 0— »2), parietal downgrowths contacing 
parabasisphenoid (67, 1.0), optic foramina enclosed in bone (68, 1.0), 
anterior brain cavity floored by frontals and cultriform process (69, 
0.67, 0->2), trigeminal foramina bordered at least partly by parietal 
(70, 1.0), supraoccipital on skull roof behind parietal (86, 0.5), 
posttemporal fenestra closed (89, 0.5), opening of Jacobson's organ 
enclosed by vomer and septomaxilla only (94, 0.4, 1— >2), vomer 
medial to palatine (99, 1.0), palatine- vomer contact mobile (102. 
1 .0), palatine long ( 1 03, 0.33, 1 — >0), palatine with distinct rectangu- 
lar process ( 105, 1 .0), two or fewer mental foramina on lateral surface 
of dentary (117, 1.0), posterior margin of lateral surface of dentary 
deeply notched (123, 0.4, l-»2), dentary overlapped by surangular 
( 1 24, 1 .0, 0— >2), surangular extends far over lateral surface of dentary 
(134, 1.0, 2— >3), articular fused with prearticular and surangular 
(144, 0.25, 1— >0), retroarticular process short (145, 1, 0— >1), mar- 
ginal teeth with medial and lateral carinae (153, 0.67, 1 — >2), palatine 
teeth present (166, 0.33, 1— >0), palatine teeth long fangs (167, 1.0), 
*at least 1 20 presacral vertebrae ( 1 80, 0.6, 3— >5), *lymphapophyses 
present (200, 0.67).*shoulder girdle absent (203, 0.67, l->2), Clavi- 
cle absent (207, 0.33), ossified sternum absent (213, 1.0), *forelimbs 
absent (218, 0.5, 1— >2), scleral ossicles absent (241, 0.5), *appen- 
dicular epiphyses absent (246, 0.67, 1— >2). 



axus^nsz-/ 



Bull. not. Hist. Mus. Loud. (Zool.) 68(2): 143-154 



Issued 28 November 2002 



Phallus morphology in caecilians (Amphibia, 
Gymnophiona) and its systematic utility 

DAVID J. GOWER AND MARK WILKINSON 

Department of Zoology, The Natural History- Museum, London SW7 5BD, UK. 
email addresses: davig@nhm.ac.uk, marw@nhm.ac.uk 

CONTENTS 

Introduction 143 

Abbreviation used in text 144 

Abbreviations used in figures 144 

Morphology 144 

Disposition of the cloaca 144 

Divisions of the cloaca 146 

Urodeum 146 

Blind sacs 146 

Anterior phallodeum 147 

Posterior phallodeum 148 

Phallodeal ornamentation 148 

Composition of phallodeal structures 148 

Relationship between the uneverted cloaca and the phallus 149 

Systematics 149 

Is phallus morphology species specific? 149 

Species differentiation and generic identity 152 

Discussion 152 

Acknowledgements 153 

References 153 



Synopsis. The cloaca of male caecilian amphibians (Gymnophiona) is a tube that comprises an anterior urodeum and a posterior 
phallodeum. The phallodeum everts (with the urodeum lying inside it) to form a phallus used for direct sperm transfer in 
copulation. Phallodeal morphology is rich in detail and variation, and has therefore been considered a potentially useful and much 
needed tool for caecilian phylogenetics and species-level taxonomy. Despite this, it has been almost entirely ignored in caecilian 
systematics. there is confusion regarding some aspects of morphology, and variation within and among species is poorly 
understood. A short review and reconsideration of phallus morphology is presented, and the systematic potential assessed. The 
anterior part of the phallodeum appears to offer the most obvious systematic potential, and the morphology of longitudinal ridges 
and their ornamentation here seem to have diagnostic and/or phylogenetic value for some taxa. Although there is evidence of 
intraspecific variation, at least some of which is associated with ontogeny and reproductive condition, individuals of the same 
species generally have a common pattern of phallodeal ridges and ornamentation, and congeners often share a similar pattern. 
However, these patterns are not universally species specific, at least among uraeotyphlids. Although variation needs to be better 
understood, the male cloaca offers great potential for caecilian systematics. 



INTRODUCTION 



As in other amphibians, caecilians (Gymnophiona) possess a cloaca, 
a chamber that opens to the exterior via the vent and into which open 
the large intestine, the urogenital (Wolffian and Miillerian) ducts, 
and the bladder. In contrast to other amphibians, the cloaca of male 
caecilians can be everted through the vent (Fig. 1) to serve as an 
intromittant organ, or phallus, used in copulation to effect direct 
sperm transfer (e.g. Himstedt, 1996). It has long been recognised 
that the external surface of the caecilian phallus and the correspond- 
ing internal surface of the uneverted cloaca may bear distinctive 
ridges and grooves, tuberosities and even spines (e.g. Duvernoy. 
1849;Gunther. 1864;Spengel, 1876; Noble, 1931). There is consid- 
erable interspecific variation in the complex patterns of these features, 
but there have been few comparative studies. 



Spengel ( 1 876) compared cloacal features in males of six species 
in what are now recognised as six genera from three families, and 
aspects of cloacal morphology were compared further in some of 
these species by Wiedersheim (1879). Tonutti (1931) provided a 
very detailed documentation of the uneverted and everted cloaca of 
the caeciliid Hypogeophis rostratus (Cuvier, 1829) and compared it 
with the uneverted cloaca of the ichthyophiid Ichthyophis glutinosus 
(Linnaeus, 1758) and of the caeciliid Spihonops annulatus (Mikan, 
1820). Tonutti (1933) expanded the comparative aspect of his study 
of the caecilian phallus by incorporating detailed data on a further 
six species, including representatives of Scolecomorphidae and 
Typhlonectidae. Tonutti's work remains the most detailed to date. 
Taylor ( 1968 and references therein) figured (though without labels 
or orientation ) everted phallodea and in situ dissections of 1 2 species 
in eight genera and four families. The broadest comparative study of 
the male cloaca was presented by Wake (1972), who examined 



© The Natural History Museum, 2002 



144 



D.J. GOWER AND M. WILKINSON 



large — 
intestine 



cloaca 



-urodeum- 



. phallodeum — 



- anterior— +*- posterior-^ 




vent 




- phallus 



Fig. 1 Schematic sagittal section through the posterior of a male 
caecilian showing (a) main divisions of the uneverted cloaca, and (b) 
the everted phallus with the internal, lumenal surface of the phallodeum 
on its exterior surface, and the urodeum forming its core. 



34 caecilian species, including representatives of 20 currently recog- 
nised genera and all six of the currently recognised families. Exbrayat 
( 1991) compared cloacae of single species from four genera in three 
families. Wake (1998) provided comparative data on the cloacal 
spines and spicules of the three nominate species of Scolecomorphus 
Boulenger, 1883. 

Species limits in caecilians are poorly understood and the tax- 
onomy within many genera is best viewed as uncertain and potentially 
unstable (Nussbaum and Wilkinson, 1989). The inadequate state of 
current knowledge has been attributed to the group's tropical distri- 
bution, largely fossorial and secretive lifestyle, under-representation 
in museum collections, lack of detailed study, and a relative paucity 
of obvious external morphological features in association with their 
limbless bodies, reduced or absent tails, and reduced head features 
(e.g. Nussbaum & Wilkinson, 1989). Some 34 years after the 
publication of Taylor's ( 1968) taxonomic monograph, species level 
caecilian systematics is still dominated by counts of annuli, verte- 
brae, and teeth. Of the phallus, Taylor ( 1968: 31) was 'certain that 
most genera and many species could be identified by the characters 
of this organ alone' and Wake ( 1972: 353) stated that 'the arrange- 
ment of musculature and cloacal accessory structures is 
species-specific in males.' If correct, male cloacal morphology, with 
its complex structure and many variations, should provide a much 
needed tool for investigating species limits in and phylogenetic 
relationships among caecilians. However, not much has changed 
since Largen etal. (1972: 187) pointed out that 'The value of penis 
structure as a taxonomic character has yet to be fully investigated'. 



We have made observations of the cloacal morphology of a broad 
range of caecilian species. Without assembling a thorough synthesis 
of these observations, we draw upon them here to provide a descrip- 
tion of the male cloaca that emphasises some features that can be 
homologised across taxa, and that indicates the kind of variation that 
occurs. It is hoped that this contribution will clarify some points of 
confusion in the literature and be a stimulus to future research. Our 
focus here is on the male cloaca only. 

Abbreviations 



UMMZ: 


University of Michigan, Museum of 


Figures 




a.ll 


anterior tuberosity of 1.1 


a.md 


anterior tuberosity of md 


ap 


anterior part of phallodeum 


a.rdl 


anterior tuberosity of r.dl 


a.rvl 


anterior tuberosity of r.vl 


b 


bladder 


bp 


blind pit 


bs 


blind sac 


c 


colliculus 


cl 


copulator loop 


c.md 


central tuberosity of md 


cs 


cloacal sheath 


ebs 


entrance to blind sac 


eu 


entrance to urodeum 


i 


intestine 


l.bs 


left blind sac 


l.dl 


left dorsolateral longitudinal ridge 


1.1 


left lateral longitudinal ridge 


l.vl 


left ventrolateral longitudinal ridge 


md 


mid-dorsal longitudinal ridge 


P 


phallodeum 


p.lvl 


posterior tuberosity of l.vl 


p.md 


posterior tuberosity of md 


PP 


posterior part of phallodeum 


p.rdl 


posterior tuberosity of r.dl 


p.rl 


posterior tuberosity of r.l 


p.rvl 


posterior tuberosity of r.vl 


r.bs 


right blind sac 


r.dl 


right dorsolateral longitudinal ridge 


r.l 


right lateral longitudinal ridge 


rm 


retractor muscle 


r.vl 


right ventrolateral longitudinal ridge 


s 


sulcus 


sph 


sphincter 


u 


urodeum 


ud 


urogenital duct 


umd 


mid-dorsal ridge of urodeum 


V 


small additional ventral tuberosity 


vd 


vent denticulations 


vp 


vascular plexus 



MORPHOLOGY 



Disposition of the cloaca. The cloaca of male caecilians is 
essentially a tube that extends between the posterior end of the 
intestines and the vent, and that may or may not have paired dorsal 
diverticula or blind sacs. The intestines, the paired urogenital ducts 



PHALLUS MORPHOLOGY IN CAECILIANS 



145 





Fig. 2 Uraeotyphlus cf. narayani (field tag MW 249). Phallodeal portion of undissected, uneverted cloaca exposed in the coelom by a mid-ventral 
incision through the body wall. The phallodeum has been rotated about its long axis through 90" to show its right lateral aspect. Scale on drawing : 
5 mm. 





Fig. 3 Uraeotyphlus cf. narayani (field tag MW 249). Anterior phallodeal portion of undissected, uneverted cloaca exposed in the coelom by a mid- 
ventral incision through the body wall. The phallodeum has been rotated about its long axis through 180° to show its dorsal aspect. See Fig. 2 for scale. 



and bladder open, in close proximity, into the cloaca at its anterior 
end. The openings of the ducts and bladder are in the dorsolateral 
and ventral wall of the cloaca respectively. The Miillerian and 
Wolffian ducts and the intestine may extend posterior to their points 
of entry into the urodeum before turning back on themselves in U- 
bends or copulator loops that facilitate the eversion of the phallus 



(Duvernoy, 1849; Gunther, 1864; Spengel, 1876; Sawaya, 1942; 
Wilkinson. 1990; this paper: Figs. 2, 3). As documented by, for 
example, Rathke (1852), Gunther (1864), Spengel (1876), 
Wiedersheim (1879: 89, Fig. 89) and Tonutti (1931, 1933: e.g. Fig. 
32), the mature male cloaca sits within a membranous cloacal 
sheath, to which it is unattached other than at its anterior and 



146 



D.J. GOWER AND M. WILKINSON 



posterior ends (e.g. Rathke, 1852; Tonutti, 1931; Exbrayat, 1996). 
This loose association presumably also facilitates cloacal eversion 
(e.g. Spengel, 1876; Wilkinson, 1990). The sheath is continuous 
with the mesorchium and with the parietal peritoneum via a ventral 
mesentary (e.g. Tonutti, 1933: Fig. 3a). 

A musculus retractor cloacae that is unique to caecilians origi- 
nates on the mid- ventral body wall and inserts posterior to its origin 
on the lateral and ventral surface of the cloaca. In those taxa 
possessing blind sacs, the insertion is bifid and is largely or perhaps 
entirely on the sacs themselves (e.g. Ichthyophis Fitzinger, 1826 
Tonutti, 1931: Fig. 30e; pers. obs.; Uraeotyphlus Peters, 1879, this 
paper: Figs. 2, 3). This muscle is thought to retract the everted 
phallodeum when contracted (e.g. Giinther, 1864; Spengel, 1876). 

Divisions of the cloaca. The cloaca can be divided along its 
long axis into two main regions (e.g. Duvernoy, 1 849; Tonutti, 193 1 ) 
- an anterior cloacal chamber, or urodeum, and a posterior cloacal 
chamber, or phallodeum (Fig. 1). The phallodeum of mature indi- 
viduals is also broadly divisible into two regions, an anterior part 
with pronounced ornamentation that forms the more distal part of 
the everted phallus, and a structurally more simple posterior section 
that forms the proximal stalk of the everted phallus. Giinther ( 1864) 
and Wiedersheim (1879) discussed three regions in the male cloaca. 
Their anterior region corresponds to the urodeum, and their middle 
and posterior parts correspond to the anterior and posterior sections 
of the phallodeum, respectively. Exbrayat ( 1 99 1 ) also distinguished 
three regions of the cloaca, but these do not correspond directly to 
the partitions recognised by other authors. His middle section 
includes the posterior part of the urodeum and the anterior 
phallodeum. 

The most obvious variations in cloacal morphology occur on the 
internal, lumenal surface of the phallodeum, which corresponds to 
the external surface of the phallus. The morphology of this surface 
can be examined directly in caecilians preserved with the phallus 
fully everted, or by dissection, serial sectioning or endoscopy 
(Himstedt, 1996). Comparison of dissected cloacae is best effected 
by maintaining an approximately standard approach. Figures of 
dissected cloacae in the literature (e.g. Duvernoy, 1849; Giinther, 
1864; Spengel, 1876; Taylor, 1968; Wake. 1972; this paper) are 
mostly of cloacae opened with a longitudinal mid- ventral incision. 
This procedure gives a clear view of the dorsal surface of the 
phallodeum. Features of the urodeum must be determined by dissec- 
tion, sectioning, or endoscopy. The caecilian phallus is sometimes 
referred to as the phallodeum (e.g. Duellman & Trueb, 1 986), but the 
latter term is more properly reserved for the posterior cloacal 
chamber. The urodeum, at least in part, also contributes to the 
phallus by forming its core as it lies inside the everted phallodeum 
(e.g. Tonutti, 1931: Fig. 22b; this paper: Fig. 1). 

In the majority of caecilians, the distinction internally between 
the urodeum and phallodeum is obvious in dissected specimens. The 
relatively simple and narrow urodeum gives way posteriorly to the 
broader phallodeum, which has pronounced longitudinal (and/or 
oblique) ridges and deep sulci extending to the phallodeal-urodeal 
border (e.g. see figures of Uraeotyphlus below). In most taxa, a mid- 
dorsal protuberance marks the posterior end of the urodeum. This 
protuberance is here termed colliculus (= little hill). The colliculus is 
perhaps equivalent, at least in part, to the 'bourrelet' mentioned by 
Duvernoy (1849; also Exbrayat, 1991). Typically the colliculus 
projects into the phallodeal chamber to a varying degree, being 
particularly large in some species (e.g. pers. obs. of Gegeneophis 
ramaswamii Taylor, 1 942, Herpele squalostoma (Stutchbury, 1 834), 
and Microcaecilia unicolor (Dumeril, 1864)). In species with blind 
sacs, these open into the phallodeum adjacent to its border with the 



urodeum. A major exception to this general pattern is apparently 
restricted to the caeciliid genera Dermophis Peters, 1879 and 
Gymnopis Peters, 1874 (MW, pers. obs.). In these caecilians, which 
lack blind sacs, there is no definite colliculus and no clear differen- 
tiation between urodeum and phallodeum. Given the apparently 
universal presence of distinct phallodeal and urodeal chambers in all 
other caecilians, including all non-caeciliids (outgroups), we inter- 
pret its absence as a putative synapomorphy of Dermophis and 
Gymnopis. 

Wake (1972) made no use of a clear urodeum-phallodeum divi- 
sion in her descriptions. She documented several features close to 
the openings of the urogenital ducts, which are in the anterior 
urodeum rather than the phallodeum. In our experience, this is a far 
more irregular region in which gross morphological regularities are 
less apparent and variation is harder to characterise than in the 
phallodeum. Wake (1972) mostly examined partially opened 
cloacae in which only the anterior part of the phallodeum could be 
observed. 

The absolute and relative sizes of the urodeum and phallodeum 
may vary taxonomically but substantial variation within species 
might be expected given that the cloaca must serve both reproduc- 
tive and alimentary functions. Exbrayat (1991) has presented 
evidence of seasonal variation correlated with the breeding cycle in 
Typhlonectes compressicauda (Dumeril and Bibron, 1841), and 
short term changes might even occur with the passage of faeces. In 
a sample of 1 1 preserved Hypogeophis rostratus, the phallodeum 
ranged from 1.6 to 5.3 times the length of the urodeum (MW, pers. 
obs.), demonstrating considerable intraspecific variation in size in 
this species. 

Urodeum. The urodeum is a relatively simple and typically nar- 
row chamber. Its dorsal surface is characterised by a pronounced 
mid-dorsal longitudinal ridge (see figures of Uraeotyphlus below) 
and seemingly irregular arrangements of other, less pronounced 
ridges. The appearance of the lesser ridges can vary substantially 
with state of preservation and possibly also in life. The colliculus is 
an expansion of the posteriormost part of the mid-dorsal urodeal 
ridge, and it shows variations in form that may be of systematic 
value, as may differences in the overall shape of the urodeum (long 
and narrow or short and somewhat broader). Additional lateral or 
ventral more pronounced longitudinal ridges may also be present in 
the urodeum (Wake, 1972). Wake (1972) described considerable 
variation in the form of the urodeum at the points of entry of the 
urogenital ducts, which are often depressed and may vary in their 
relations to the mid-dorsal longitudinal ridge. She reported that 
papillae associated with the openings of the urogenital ducts were 
present only the typhlonectids {Typhlonectes compressicauda, 
Chthonerpeton indistinctum (Reinhardt and Liitken, 1861) and C. 
viviparum Parker and Wettstein, 1 929) that she examined. However, 
one of us (MW) has observed urogenital papillae in other species, 
including taxa that Wake reported as lacking them (e.g. Grandisonia 
sechellensis (Boulenger. 1 909) and Geotrypetes seraphini (Dumeril, 
1859)). Systematically useful variation may occur in the urodeum 
but we have not yet discerned clear patterns of variation. 

Blind SACS. Blind sacs (caecal appendage of Giinther, 1864; 
Penisblindsack of Spengel, 1876; Blindsack of Wiedersheim, 1879; 
Penissack of Tonutti, 1931) are paired anterior extensions of the 
phallodeum that run parallel to the urodeum (Figs. 2, 3). Blind sacs 
vary in size and they may be free or partially fused to the adjacent 
urodeum (e.g. Wake, 1972). In species with blind sacs, these are a 
feature of the mature cloaca and may be absent or less well devel- 
oped in immature males (see discussion of Uraeotyphlus below). In 
most cases, species within the same genus, or that are otherwise 



PHALLUS MORPHOLOGY IN CAECILIANS 



147 




Fig. 4 Uraeotyphlus cf. narayani ( Held tag MW 207). Views of (a) right lateral, (b) dorsal, (c) distal and slightly ventral, and (b) ventral surfaces of 
phallus (everted cloaca). Scale bar for Fig. 4b = 3 mm. 



considered to be closely related, have blind sacs in a similar con- 
dition, suggesting relatively stable and systematically informative 
interspecific variation. Blind sacs are well developed in ichthyophiids 
and uraeotyphlids, caecilians that Wake (1972) considered 'primi- 
tive' in other reproductive characters, leading her to suggest that 
well developed blind sacs are a general caecilian feature, with 
reduction and loss being derived. In contrast, Tonutti (1931. 1933) 
considered well developed blind sacs derived. Rhinatrematids are 
believed to be the sister group of other extant caecilians on the basis 
of a wide variety of evidence (e.g. Nussbaum, 1977; Hedges et <//., 
1993; Wilkinson, 1996). Spengel (1876) and Wake (1972) docu- 
mented blind sacs in the rhinatrematids Rhinatrema bivittatum 
(Cuvier, 1829) and Epicrionops petersi Taylor, 1968 respectively, 
but we note their absence (or minimal development) in mature 
Epicrionops marmoratus Taylor, 1968 (MW, pers. obs.). This sug- 
gests homoplasy and may complicate the interpretation of polarity. 

Anterior phallodeum. The lumenal surface of the anterior 
phallodeum bears the distinctive structures seen on the external 
surface of the more distal part of the fully everted phallus (Figs. 1,4 
to 9). Much variation occurs here, but we discern a presumably 
homologous pattern anteriorly that is common to almost all caecilians. 
In this region there is a pair of deep dorsolateral grooves, one on 
either side. Each of these sulci (Figs. 4 to 9) are bordered by a pair 



of well developed, parallel dorsolateral longitudinal or oblique 
ridges. A median mid-dorsal longitudinal ridge may or may not also 
be present, a variation that appears to be species specific. In species 
with blind sacs, the sulci and their bordering ridges run into the blind 
sacs, extending to their distal tips. In species lacking blind sacs, the 
ridges fade out and the sulci open out at the anterior of the phallodeum, 
either side of the colliculus. In Hypogeophis rostratus. the sulci run 
posteriorly and terminate blindly with the fusion of their associated 
ridges (Tonutti, 193 1 : Fig. 20; pers. obs.), a pattern that is consistent 
in the 1 1 specimens of this species examined by one of us (MW). 
Similar 'fusion' of the dorsolateral longitudinal ridges occurs in 
many caecilians (e.g. Uraeotyphlus, Figs. 6 to 9). Less commonly, 
the posterior end of each sulcus is open, with the more medial 
bordering ridge fading out or fusing with its antimere along the 
dorsal midline (e.g. Grandisonia alternans (Stejneger, 1893), 
Gegeneophis ramaswamii, Boulengeritla boulengeri Tornier, 1896, 
MW, pers. obs.). Additional major longitudinal ridges may or may 
not be present lateral and/or ventral to those forming the sulci. In 
uraeotyphlids (Figs. 4 to 9) and ichthyophiids, major longitudinal 
ridges are broadly distributed, whereas in some caeciliids (pers. obs. 
of e.g. Grandisonia Taylor, 1 968 and Schistometopum Parker, 1941; 
this paper; Fig. 10) the ridges are more restricted to the dorsal 
surface of the phallodeum. Although we have discussed a single 
main pair of sulci, there may be other, smaller, more or less 



148 



D.J. GOWER AND M. WILKINSON 





Fig. 5 Uraeotyphlus cf. narayani (field tag MW 207). Views of (a) dorsal, and (b) distal and slightly ventral surfaces of phallus (everted cloaca). For 
scale see Fig. 4. 



rm 



a.rvl 




Fig. 6 Uraeotyphlus cf. narayani (field tag MW 254). Dissected cloaca 
of mature male. The cloaca has been opened mid-ventrally and pinned 
to reveal the lumenal surface of the phallodeum and posterior part of the 
urodeum. The incision has longitudinally bisected the right ventrolateral 
longitudinal ridge so that parts of it lie on each side of the open cloaca. 
Scale = 3 mm. 



longitudinal grooves at the anterior end of the phallodeum, at least 
some of which may enter the blind sacs, where present (e.g. 
Geotrypeles Peters, 1879, pers. obs.). 

POSTERIOR PHALLODEUM. The distinction between the anterior 
and posterior phallodeum is sometimes less clear cut than that 
between the phallodeum and urodeum. Wake (1972) reported that 
the longitudinal ridges of the anterior phallodeum continue 
posteriorly to the vent. We find that the major longitudinal ridges 
reduce greatly posteriorly, either abruptly or gradually, that they 
may or may not extend as far as the vent, and that the pattern of 
ridges within the posterior phallodeum is irregular or less obviously 
regular than those of the anterior phallodeum. The phallodeum 
narrows dramatically posteriorly, shows considerable variation in 
length, and has its terminal portion surrounded by a sphincter of 
variable size. 

PHALLODEAL ORNAMENTATION. The major longitudinal ridges 
of the anterior phallodeum may be more or less invested with, or 
elaborated into, tuberosities, transverse ridges and grooves, longi- 
tudinal crests, or spines that are often in distinctive patterns (e.g. 
Figs. 6, 9). Isolated thickenings or other ornaments may also 
occur in the spaces between the major longitudinal ridges. The 
ridges associated with the dorsolateral sulci bear such features 
only posterior to the sulci (e.g. Figs. 4, 7, 9). Both the shape and 
arrangement of this ornamentation may be expected to provide 
systematic characters, although there is also evidence of 
intraspecific variation (e.g. Scolecomorphus, Wake, 1998). Species 
appear to differ in whether the ridges within the posterior 
phallodeum bear any ornamentation or not. Where present, as in 
Typhlonectes compressicauda (Exbrayat 1996), they are not as 
pronounced or distinctive as the structures of the anterior 
phallodeum (distal phallus). 

Composition of phallodeal structures. The composition of 
the main longitudinal ridges and their ornamentation is unclear from 



PHALLUS MORPHOLOGY IN CAECILIANS 



149 



the literature and warrants further histological examination. Tonutti 
( 1 93 1 , 1 933) viewed the longitudinal ridges as encompassing longi- 
tudinal 'propulsor' muscles but we are unable to verify this from his 
figured sections. Wake (1972: 354) described the ridges as 'longitu- 
dinal muscles overlain by fibrous connective tissue', but also warned 
(p. 363) that 'Caution must be exercised in interpreting the various 
folds in the cloacal wall. They may often not be muscle but may be 
ridges of connective tissue'. Wake (1998) referred to connective 
tissue ridges in Scolecomorphus and made no mention of previous 
reports that ridges are muscular (Tonutti, 1933; Wake, 1972). Wake 
(1972) also referred to at least some phallodeal ornamentation as 
transverse muscle ridges, whereas Wiedersheim (1879) stated that 
the prominences are hardened parts of longitudinal folds of cloacal 
mucosa. In at least one case it is clear that the prominences are not 
muscular: large recurved calcified or cartilaginous spines are present 
in Scolecomorphus uluguruensis Barbour and Loveridge, 1925 
(Noble, 1931; Taylor, 1968;Nussbaum, 1985; Wake, 1998).Exbrayat 
(1991) showed that tuberosities in the phallodeum of Typhlonectes 
compressicauda are keratinous, and that their thickness varies with 
the reproductive cycle. Exbrayat (1996) described smooth trans- 
verse and striated longitudinal muscles in the wall of the cloaca of T. 
compressicauda, with the latter forming the major longitudinal 
ridges. Muscle therefore appears to be present in the longitudinal 
phallodeal ridges of at least some species, but we find no clear 
evidence that any of the tuberosities, crests etc found in the 
phallodeum are muscular. 



close to the border between the phallodeum and urodeum. The sacs 
extend anteriorly from the anterior end of the phallodeum so that, 
within the coelom, they can be seen running parallel to the posterior 
end of the urodeum (e.g. Wiedersheim, 1879: Fig. 88; this paper: 
Figs. 2, 3). Thus, the blind sacs must be positioned at, or inside, the 
distal end of the everted phallus (Tonutti, 1931: e.g. Fig. 22b of 
Hypogeophis rostratus) rather than at its base. This can be clearly 
seen by comparing the figures shown here of the uneverted and 
everted phallodeum of Uraeotyphlus (Figs. 2 to 9), where the 
entrance to the blind sacs are seen right at the distal termination of 
the everted phallus (Figs. 4, 5). Preserved specimens may show 
various degrees of phallodeal eversion, and it is clear that Wake's 
figures are of partially everted organs, which may have misled her. 
In our experience, the major dorsolateral sulci, their associated 
ridges, and the colliculus are clearly visible at the distal end of a well 
everted phallus, although the extent of phallodeal eversion during 
copulation is unknown. 

Bons (1986) and Exbrayat (1991) also figured what we consider 
to be partially everted phallodea of Typhlonectes compressicauda. 
Typhlonectes have a distinctive 'cloacal disc' surrounding the vent 
(Taylor, 1968) and Exbrayat's figure 3 appears to show the cloacal 
disk at the distal tip of the protruding phallus, and seemingly 
detached from the adjacent skin. However, the disc is continuous 
with the surrounding skin and must remain at the base of the phallus 
because it is everted rather than telescopically extended. 



Relationship between the uneverted cloaca and the phallus. 
There is some confusion in the literature regarding the positional 
relationship between structures as seen on the internal surface of the 
uneverted phallodeum. and the same structures when observed on 
the external surface of the phallus. Wake (1972: 359, Fig. 13, 15) 
described and figured the blind sacs as being positioned at the 
proximal base of the everted phallodeum in a thickened 'blind sac 
sheath'. In the uneverted phallodeum, blind sacs, where present, are 
pockets extending from the dorsal wall of the phallodeum, very 



SYSTEMATICS 



IS PHALLUS MORPHOLOGY SPECIES SPECIFIC? The family 

Uraeotyphlidae is monotypic, comprising five currently recognised 
species of Uraeotyphlus endemic to peninsular India (Pillai & 
Ravichandran, 1999). Uraeotyphlidae is the extant sister taxon of 
the south and southeast Asian Ichthyophiidae (Wilkinson & 
Nussbaum, 1996; Gower et «/., 2002; Wilkinson et al., 2002). As 





c.md 



Fig. 7 Uraeotyphlus cf. narayani (field tag MW 172). Anterior phallodeum of mature male, prepared as specimen shown in Fig. 6. Scale = 2 mm. 



150 



D.J. GOWER AND M. WILKINSON 



>J 





Fig. 8 Uraeotyphlus cf. narayani (UMMZ 139810). Cloaca of immature male, prepared as specimen shown in Fig. 6. Scale = 2 mm. 



with many groups of caecilians, the taxonomy of Uraeotyphlus has 
an inadequate basis, with some species known from only few 
specimens, many with poor locality data. Few diagnostic characters 
have been identified and current keys are not satisfactory, so that 
caution needs to be exercised in applying names to individuals, and 
in assuming species identity of groups of individuals. The following 
discussion draws on the examination of the cloaca in more than 30 
male Uraeotyphlus representing at least three distinct species. The 
focus here is on features of the lumenal surface of the anterior 
portion of the phallodeum, chiefly the longitudinal ridges and their 
ornamentation. 

Figures 4 to 8 show the morphology of the phallus and dissected 
cloacae of four specimens. These are identified as Uraeotyphlus cf. 
narayani Seshachar, 1 939, but unpublished morphological and mole- 
cular data have revealed previously unsuspected diversity in the 
populations that these individuals are drawn from. It is not yet 
apparent whether this diversity is indicative of previously unrecog- 
nised specific or subspecific taxa. Whatever their true specific 
identity, these four specimens share a common pattern in the major 
features of the anterior phallodeum. There are seven major longitu- 
dinal phallodeal ridges - a single mid-dorsal ridge, and pairs of 
dorsolateral, lateral, and ventrolateral ridges. As in most other 
caecilians, the anterior end of each dorsolateral ridge holds a major 
longitudinal sulcus that extends into the corresponding blind sac 
(Figs. 4, 5, 9). In mature individuals, each of the major longitudinal 
ridges bear hardened transverse thickenings. When relatively small, 
these thickenings bear an approximately transverse narrow line of 
dense, opaque tissue that stands out against the more translucent 
main body of longitudinal ridge. Where relatively large, the 
thickenings are developed into tuberosities that can be irregular, and 
that interlock in the uneverted cloaca. The mid-dorsal ridge bears 
three such tuberosities and the other, paired longitudinal ridges two 
each. The transverse thickenings of each major longitudinal ridge 
are offset relative to each adjacent ridge, and they generally bear the 
same spatial relationship to each other in each individual (Figs. 4 to 
7). Of the paired ridges, the lateral ones are the least well developed, 



and sometimes they are best located by their transverse thickenings. 
Within this common pattern are some minor variations. In immature 
males (Fig. 8), the main longitudinal ridges are less well developed 
and bear no transverse thickenings or indications of hardened tissue, 
but they can still be readily identified and homologised with those in 
mature males. In addition, the blind sacs of immature males are not 
developed. Instead, there is a pair of shallow pits in their place. The 
relative size of the transverse thickenings or tuberosities also varies 




Fig. 9 Uraeotyphlus cf. oxyurus (field tag MW 469). Cloaca of mature 
male prepared as specimen shown in Fig. 6. The incision has 
longitudinally bisected the left ventrolateral longitudinal ridge so that 
parts of it lie on each side of the opened cloaca. The left side of the 
posterior end of the urodeum has been torn away from the anterior end 
of the phallodeum so that retractor muscle is visible through the 
resulting hole. Scale = 3 mm. 



PHALLUS MORPHOLOGY IN CAECILIANS 



151 




Fig. 10 Schistometopum gregorii from Tanzania. Views of (a) dorsal, and (b) right lateral surface of phallus of field specimen MW 3257, and (c) dorsal, 
and (d) ventral surface of phallus of field specimen MW 3251. Scale bars in mm. 



152 



D.J. GOWER AND M. WILKINSON 



among individuals, but whether this variation is correlated with 
taxonomy, ontogeny, and/or temporally within any possible repro- 
ductive cycles is as yet unknown. Occasionally, minor variations in 
the ornamentation are seen. For example, the individual shown in 
Fig. 4 also has a single, poorly formed, transverse thickening 
ventrally. In the individual shown in Fig. 6, the posteriormost 
transverse thickening on the right dorsolateral longitudinal ridge 
extends posterior to the posteriormost transverse thickening on the 
mid-dorsal longitudinal ridge, whereas the reverse of this pattern (as 
seen on the left of this individual) is more commonly encountered. 
Finally, the transverse thickenings or tuberosities are sometimes 
multipartite. 

Figure 9 depicts the phallodeum of an individual identified as U. 
cf. oxyurus (Dumeril and Bibron, 1841). Although the precise 
specific identity of this individual also is not entirely clear, we are 
confident that it is referable to a species distinct from that (or those) 
represented in Figs. 4 to 8. For example, the U. cf. oxyurus indi- 
vidual comes from a population with substantially more vertebrae 
(112-115, n = 18) than the populations represented by the other 
figured specimens (93-1 10, n > 100). Despite their apparent specific 
distinctness, the phallodea of U. cf. narayani (Figs. 4 to 8) and U. cf. 
oxyurus (Fig. 9) share the same number and pattern of longitudinal 
ridges and transverse ornamentation. Thus Wake's (1972: 353) 
claim that the phallodeal ridges and 'cloacal accessory structures is 
species-specific' does not appear to hold - at least not at the level of 
the presence, number, or topographical relations of major features. It 
might yet hold for morphometric variations of phallodeal features 
and/or for fine morphological details of the longitudinal ridges and 
their ornamentation, but this needs further assessment. 

That not all species of Uraeotyphlus share the same basic 
phallodeal morphology is revealed by observation of U. cf. 
malabaricus (Beddome, 1870), in which the number and arrange- 
ment of longitudinal ridges and their ornamentation is markedly 
different. Interestingly, analysis of mitochondrial DNA sequence 
data strongly indicates that U. narayani and U. cf. oxyurus share a 
more recent common ancestor with each other than either does with 
U. cf. malabaricus (Gower et al, 2002). 

Species' differentiation and generic identity. Nussbaum & 
Pfrender's (1998) recent revision of the caeciliid genus 
Schistometopum recognised two species occurring on opposite sides 
of the African continent. S. thomense (Barboza du Bocage, 1873) is 
known from Sao Tome island in the Gulf of Guinea, and S. gregorii 
(Boulenger, 1894) from lowland coastal regions of Kenya and 
Tanzania. The validity of the genus has not been seriously ques- 
tioned, but it is currently diagnosed on a combination of characters, 
with no known unique synapomorphies. 

Wake (1972: 358) described the male cloaca of S. thomense as 
having 'four regularly spaced muscle bands on each side of the 
cloaca', presumably features of the urodeum, and that 'the posterior 
part of the cloaca [more the central region, as can be seen when the 
cloaca is fully dissected] is arranged in three sets of transverse, 
crescent-shaped muscles, one mid-dorsal, the other two ventro- 
lateral.' Tonutti (1933) described longitudinal phallodeal ridges as 
dorsal rather than ventrolateral in 5. thomense and we concur with 
his assessment (see Fig. 10). Wake found the cloaca of S. gregorii to 
have a similar morphology to that of S. thomense. Although we are 
not convinced that the transverse ridges comprise muscle, we agree 
that the two species share a similar morphology, and consider the 
presence of three (though see discussion of S. gregorii below) 
narrow and long longitudinal ridges with a characteristic ornamen- 
tation of regularly spaced, scalloped transverse ridges and grooves 
to be restricted to these two species among material we have 





a 



Fig. 11 Sketches showing disposition of major longitudinal ridges and 
their ornamentation in the dorsal lumenal wall of the anterior part of the 
phallodeum of (a) Schistometopum thomense (UMMZ 188027), and (b) 
S. gregorii (UMMZ 14701 1 ) from Kenya. Compare with Tanzanian 5. 
gregorii shown in Fig. 10. Not drawn to scale. 

observed. Thus, this phallodeal structure is potentially a unique 
diagnostic character of Schistometopum. 

Wake (1972) considered the phallodeal ridges of Schistometopum 
to resemble the condition in Geotnpetes. However, the part of the 
mid-dorsal longitudinal ridge that bears ornamentation in both S. 
thomense (Fig. 1 la) and S. gregorii (Figs. 10, lib) is relatively much 
longer than the comparable ornamented area in Geotrypetes 
seraphini, which is instead restricted to a small nubbin that lies at, or 
slightly beyond, the level of the posterior end of the ornamented part 
of the longitudinal ridges lateral to it (pers. obs. of e.g. UMMZ 
172648). In addition, the ornamentation appears to be somewhat 
different in the two genera, which otherwise also have quite differ- 
ently organised cloaca (for example, Schistometopum lacks blind 
sacs). 

The phallodeum of a single specimen (UMMZ 147011) of S. 
gregorii from Northern Kenya has been examined and a sketch of 
the ornamented part of the longitudinal ridges is shown in Fig. 1 lb. 
The figured morphology is largely similar to that seen in several 
specimens of S. thomense (e.g. Fig. 11a), except that, in UMMZ 
147011, there is not a single mid-dorsal ridge, but instead two 
paramedian longitudinal ridges, one longer than the other. Both of 
these ridges bear transverse crests, but they are shorter relative to the 
dorsolateral longitudinal ridges than in the observed specimens of S. 
thomense. The morphology of the mid-dorsal region of the phallo- 
deum in two Tanzanian specimens of S. gregorii observed for this 
study (Fig. 10) both bear a greater resemblance to the condition in S. 
thomense (Fig. 11a) than to the single Kenyan 5. gregorii (Fig. 1 lb) 
examined. The sample size is small, but the observed morphological 
variation is intriguing in light of Taylor's (1968: 677) suggestion 
that, based on differences in annulation, the Tanzanian and Kenyan 
populations of S. gregorii might be specifically distinct. 



DISCUSSION 

The complex structure of the caecilian phallus offers great potential 
for caecilian systematics, both as a source of diagnostic features for 
species, and of characters for phylogenetics. However, to fully 
exploit this potential requires a better understanding of the extent of 
intraspecific variation that occurs within features that appear to vary 
interspecifically. Of course, in this regard there is no difference 
between the caecilian phallus and any other structure employed in 
systematics, and we suggest that incomplete understanding of vari- 
ation should temper but not discourage the use of cloacal characters 



PHALLUS MORPHOLOGY IN CAECILIANS 



153 



in caecilian systematics. There is evidence of considerable ontoge- 
netic variation in the development of blind sacs and phallodeal 
ornamentation, emphasising the need for systematic comparisons to 
be of co-ordinate developmental stages or of developmental trajec- 
tories. There is also evidence of variation in adults in the sizes of the 
urodeum and phallodeum, and the exact form of ridges, their orna- 
mentation, and other phallodeal structures, at least some of which is 
seemingly correlated with breeding cycles. Despite Wake's (1998: 
183) statement that the morphology of the phallodeum of 
Scolecomorphus 'is indeed consistent within the species', the same 
paper clearly documents intraspecific variation in the number of 
phallodeal spines in Scolecomorphus uluguruensis and S. villains 
(Boulenger, 1895). Functional considerations lead us to speculate 
that additional intraspecific variation in phallodeal ornamentation 
occurs because the phallodeum serves both reproductive and excre- 
tory roles. In individuals with well-developed tuberosities, these can 
interdigitate in situ to seemingly obstruct the cloacal lumen. We 
hypothesise that in these species, at least, cloacal ornamentation 
would be elaborated at times of courtship but reduced at other times. 
If correct, differences in reproductive condition would need to be 
taken into account in any systematic comparisons. 

Our observations suggest that the pattern of major longitudinal 
ridges and often also the number and position of phallodeal tuberosi- 
ties or other ornamentation is mostly constant within species. The 
same general pattern occurs in 1 1 specimens of Hypogeophis 
rostratus, the largest sample of a single species that we have 
examined in detail. However, detailed study of ontogenetic and 
population variation is needed to test this constancy and to deter- 
mine whether variations in the form of phallodeal ornamentation are 
of systematic utility. Thus, future studies should attempt to increase 
sample sizes for at least some species. Of the 33 species examined by 
Wake (1972), her largest sample was 29 specimens of Gymnopis 
proximo (Cope, 1877) whereas sample sizes for the remaining 
species were low (mean = 1.7), providing little basis for assessing 
variation. Wake ( 1 972) did not discuss intraspecific variation in any 
species. 

Closely related species (e.g. congeners) tend to have similar 
cloacal morphologies, providing a strong indication that the cloaca 
will be a source of stable phylogenetic characters. For example, the 
absence of a definitive colliculus or any other obvious division of the 
cloaca into urodeal and phallodeal chambers is a very striking 
putative synapomorphy of Dermophis and Gymnopis. These genera 
have been considered closely related (e.g. Nussbaum & Wilkinson. 
1989) but there are no previously reported uniquely derived 
characters. Similarly, the general form of the longitudinal phallodeal 
ridges and their ornamentation in Schistometopum thomense and S. 
gregorii appears to offer the first known unique diagnostic character 
for Schistometopum. On the other hand, congeners can sometimes 
be readily distinguished by clear-cut, discrete differences in the 
patterns of phallodeal ridges and topological relations in their 
ornamentation. 

Contrary to Wake (1972), our investigations of Uraeotyphlus 
suggest that, in at least some cases, cloacal morphology may not be 
species specific. Instead, it appears that some species that can be 
clearly differentiated based on traditional morphological characters 
have a common pattern of phallodeal ridges and ornamentation. 
Species specific differences in these examples may yet be found in 
the details of the form of phallodeal morphology, but additional 
work is needed to test this. 

In this survey we have concentrated upon the gross structural 
features of the caecilian cloaca. The lumenal surface of the cloaca 
appears to be also covered in many minor ridges and grooves 
(striae). This micro-ornamentation may also yield useful systematic 



data but, as with more macroscopic features, studies of this must 
take into account potential intraspecific variation. In some cases, 
where we have described major structures as terminating, it might be 
more accurate to describe them as giving rise to, or being supplanted 
by, striae. For example, in Hypogeophis rostratus, where the main 
dorsolateral longitudinal ridges and their sulci 'terminate' anteriorly, 
close to the colliculus, they more accurately continue into incon- 
spicuous striae (MW, pers. obs.). These bend around the lateral 
margins of the colliculus and open into channels running alongside 
the main mid-dorsal urodeal ridge. We suspect this arrangement 
constitutes the passage through which sperm travel from the urodeum 
to the phallodeum, to be delivered to the female via the dorsolateral 
sulci that are such a prominent feature of the phallus. 



Acknowledgements. It is a pleasure to thank our esteemed colleague 
and friend Garth Underwood for inspiring our research and for enlightening 
discussions of systematics, cloacae and histology. The assistance in provision 
of material of too many people to name individually has been indispensable 
to this work, and is gratefully acknowledged. Thanks to Christian Klug and 
Claudine Levasseur for assistance with translations, Alex Kupfer for helpful 
discussions of cloacal form and function, and Harry Taylor for Fig. 10. MW 
is grateful to Ron Nussbaum for fostering and encouraging his interest in 
caecilian cloacae, and to David Sever for helpful discussion of amphibian 
urogenital systems. This paper was improved by critiques from Barry Clarke. 
Alex Kupfer, Simon Loader and Hendrik Miiller and supported in part by 
NERC "rants GST/02/832 and GR/9/2881. and an MRF award. 



REFERENCES 



Bons, J. 1986. Donnees histologiques sur le tube digestif de Typhlonectes 

compressicaudus (Dumeril et Bibron. 1841) (amphibien apode). Memoires de la 

Sociele Zoologique de France 43: 87-90. 
Duellman. W. E. & Trueb, L. 198ft. Biology of amphibians. New York, 670pp. 
Duvernoy, G. L. 1849. Cours d'histoire naturelle des corns organises professe au 

college de France. Revue et magasin de Zoologie 1849: 179-189. 
Exbrayat, J-M. 1991. Anatomie du cloaque chez quelques gymnophiones. Bulletin de 

la Soncie Herpetologique de Frame 58: 31-43. 
1996. Croissanceet cycle du cloaque chez Typhlonectes compressicaudus (Dumeril 

et Bibron. 1 84 1 ). amphibien gymnophione. Bulletin de la Sociele Herpetologique de 

trance 121: 93-98 
Gower, D. J., Kupfer, A., Oommen, O. V., Himstedt, W., Nussbaum, R. A., Loader, 

S. P., Presswell, B., Miiller, H., Krishna, S. B., Boistel, R. & Wilkinson, M. 2002. 

A molecular phylogeny of ichthyophiid caecilians (Amphibia: Gymnophiona: 

Ichthyophiidae): Out of India or out of southeast Asia? Procceedings of the Royal 

Society B 269: 1563-1569. 
Giinther, A. C. L. G. 1864. Reptiles oj British India. London. 452pp. 
Hedges, S. B., Nussbaum, R. A., & Maxson, L. R. 1993. Caecilian phylogeny and 

biogeography inferred from mitochondrial DNA sequences of the I2S rRNA and 

16S rRNA genes (Amphibia: Gymnophiona). Herpetological Monagaphs 7: 64-76. 
Himstedt, W. 1996. Die Blindwiihlen. Magdeburg, 160pp. 
Largen, M. J., Morris, P. A. & Yalden, D. W. 1972. Observations on the caecilian 

Geotrypetes grandisonae Taylor (Amphibia: Gymnophiona) from Ethiopia. Monitore 

Zoologico Italiano, Supplemento IV 8: 185-205. 
Noble, G. K. 1931. The biology of the Amphibia. New York, 577pp. 
Nussbaum, R. A. 1977. Rhinatrematidae: a new family of caecilians (Amphibia: 

Gymnophiona). Occasional Papers of the Museum of Zoology, University of Michi- 
gan 682: 1-30. 

1985. Systematics of Caecilians (Amphibia: Gymnophiona) of the family 

Scolecomorphidae. Occasional Papers of the Museum of Zoology. University of 
Michigan 713: 1-19. 

& Pfrender, M. E. 1 998. Revision of the African caecilian genus Schistometopum 

Parker (Amphibia: Gymnophiona: Caeciliidae). Miscellaneous Publications, Mu- 
seum of Zoology, University of Michigan 187: 1-32. 

& Wilkinson, M. 1989. On the classification and phylogeny of caecilians 

(Amphibia: Gymnophiona). a critical review. Herpetological Monogaphs 3: 1-42. 

Pillai. R. S. & Ravichandran, M. S. 1999. Gymnophiona (Amphibia) of India: a 
taxonomic study. Records of the Zoological Survey of India, Occasional Paper 172: 
1-117 



154 



D.J. GOWER AND M. WILKINSON 



Rathke, H. 1852. Bemerkungen uber mehrere Korpertheile der Coecilia annulate 

Archiv fur Anatomie, Physiologie und Wissenschaftliche Medicin 1852: 334—350. 
Sawaya, P. 1942. Sobre a cloaca des Siphonops. Universidade de Sao Paulo, Boletins 

da Faculdade de Filosofia, Ciencias e Letras 25 (Zoologia no. 6): 3-55. 
Spengel, J. W. 1876. Das Urogenitalsystem der Amphibien. I. Theil. Der anatomische 

Bau des Urogenitalsystems. Arbeiten aus dem zoolog.-zootom. Institut in Wurzburg 

3:51-114. 
Taylor, E. H. 1968. The caecilians of the World. Lawrence, 848pp. 
Tonutti, E. 1931. Beitrag zur Kenntnis der Gymnophionen. XV. Das Genitalsystem. 

Morphologisches Jahrbuch 68: 151-292. 
1933. Beitrag zur Kenntnis der Gymnophionen. XIX. Kopulationsorgane bei 

weiteren Gymnophionenarten. Morphologisches Jahrbuch 72: 155-211. 
Wake, M. H. 1972. Evolutionary Morphology of the caecilian urogenital system. IV. 



The cloaca. Journal of Morphology 136: 353-366. 

1998. Cartilage in the cloaca: Phallodeal spicules in caecilians (Amphibia: 

Gymnophiona. Journal of Morphology 237: 177-186. 

Wiedersheim, R. 1879. Die Anatomie der Gymnophionen. Jena, 101pp. 
Wilkinson, M. 1990. The presence of a Musculus Retractor Cloacae in female 
caecilians (Amphibia: Gymnophiona). Amphibia-Reptilia 11: 300-304. 

1996. The heart and aortic arches of rhinatrematid caecilians (Amphibia: 

Gymnophiona). The Zoological Journal of the Linnean Society 118: 135-150. 

& Nussbaum, R. A. 1996. On the phylogenetic position of the Uraeotyphlidae 

(Amphibia: Gymnophiona). Copeia 1996: 550-562. 

, Sheps, J. A., Oommen, O. V. & Cohen, B. L. 2002. Phylogenetic Relationships 

of Indian caecilians (Amphibia: Gymnophiona) inferred from mitochondrial rRNA 
gene sequences. Molecular Phylogenetics and Evolution 23: 401-407. 



Bull. not. Hist. Mus. Land. (Zool.) 68(2): 155-163 



Issued 28 November 2002 



Holaspis, a lizard that glided by accident: 
mosaics of cooption and adaptation in a 
tropical forest lacertid (Reptilia, Lacertidae) 



E.N. ARNOLD 

Department of Zoology, The Natural History Museum, Cromwell Road, London SW7 5BD. 



SYNOPSIS. Holaspis is the most morphologically apomorphic lacertid taxon with 42 or more derived morphological features 
arising on its exclusive lineage. Nearly all of these confer advantages in three specialised activities, or ameliorate problems 
resulting from them. The activities are: climbing on the often vertical open surfaces on tree boles and branches, utilising very 
narrow crevices in wood and beneath bark, and the ability, unique among lacertids. to glide from tree to tree. Although many of 
the features related to these activities are likely to result from direct adaptation to the situations concerned, exaptation has been 
critical in the development of gliding. Two behaviours present in the earliest lacertids have been coopted to this activity: rib 
spreading associated with basking contributes to an effective aerofoil, and balance control associated with running helps maintain 
appropriate posture in the air. Features originally developed in the context of crevice use also contribute to the aerofoil and a high 
surface: weight ratio. So, while natural selection has moulded Holaspis for its present activities, multiple accidents of history have 
also been important, as they also have in the evolution of bird flight. 



INTRODUCTION 



Sometimes there has been a flurry of adaptation on a lineage after a 
long period of little or no obvious change. A plethora of apomorphies 
may have been produced, often in association with shift into a new 
and demanding niche or a succession of these. For instance, this 
occurred in the lacertid lizard genus Meroles where apomorphies 
accumulated in a series of increasingly extreme soft-sand environ- 
ments (Arnold 1990, 1991 ). In other cases, not all the features that 
confer performance advantage in such a selective regime necessar- 
ily arose by natural selection in its context. In some instances, 
features developed by natural selection in a different situation or by 
some other means, and were only later coopted to a new function. 
Darwin ( 1 872) was aware of this process which was named exaptation 
by Gould & Vrba (1982). Cases of exaptation are very widespread 
(Arnold, 1994; Gould, 2002) and contribute to the ability of lineages 
to enter new selective regimes. Usually, optimum survival in these 
involves combining exaptations with new features that are built by 
the new selective regime. Exaptations are typically a small propor- 
tion of the necessary features, but there are examples where a 
number of characters really critical to invading the new regime are 
exaptations. A case in point is the aberrant lacertid genus, Holaspis, 
the only member of the approximately 1 700 species of Scincomorpha 
known to glide regularly and effectively. 



TAXONOMY AND RELATIONSHIPS 

Until recently Holaspis was regarded as a single species with two 
well-defined subspecies, but these are now each given species 
status (Broadley, 2000) as Holaspis guentheri and H. laevis. H. 
guentheri occurs in Sierra Leone, Ghana, Nigeria, Cameroon, 
Congo, Uganda, Gabon and Angola, and H. laevis in Tanzania, 
southeast Congo, Malawi and Mozambique. H. laevis has six dark 
longitudinal stripes on the body instead of eight and has on aver- 
age fewer, larger scales comprising the semitransparent window 



present in the lower eyelid which is generally rather better devel- 
oped than in H. guentheri. 

Within the Lacertidae, morphology indicates that Holaspis is a 
member of the subfamily Eremiainae (Harris, Arnold & Thomas, 
1998) and within this of the Equatorial African clade (Arnold, 
1989a, b.), which is relatively basal and has a generally primitive 
morphology, most of its members not differing much from members 
of the generally primitive subfamilies, Lacertinae and Gallotiinae. 
The Equatorial African clade is characterised by the following 
combination of derived features: pineal foramen lacking; medial 
area of the clavicle not markedly expanded; only one postnasal 
scale; parietal scale extending laterally to the edge of parietal table 
of the skull, and the tympanic scale small. All except Holaspis also 
have the postorbital and postfrontal bones fused, the absence of this 
condition in Holaspis being secondary (Arnold, 1989a). Among the 
Equatorial clade. morphology suggests Holaspis is the sister group 
of two species of Adolf us, A. africanus and A. vauereselli (Arnold 
1989a). Studies of mitochondrial DNA sequence (Harris & Arnold, 
pers. obs) corroborate this relationship, although with only low 
bootstrap support. 



MORPHOLOGY OF HOLASPIS 

The following account concentrates on those characters that are 
peculiar to Holaspis and derived within the Equatorial African 
group, and usually within the Lacertidae as a whole. These 
autapomorphies and are listed in Appendix 1. For illustrations of 
living Holaspis see Schmidt, 1919 (reproduced in Arnold 1989a), 
Schiotz (1960) and Branch (1998). 

Holaspis are small lizards growing to a maximum of only about 
53mm from snout to vent and a total length of 130mm. The whole 
animal is extremely depressed, and more so than any other lacertid. 
The index, head depth/ head width, measured on alcohol-preserved 
specimens somewhat exceeds that found in most other flattened 
lacertids (see for instance Arnold 1998a, p. 344), averaging about 
0.54 when measured in adults (n = 10). This however does not give 



© The Natural History Museum. 2002 



156 



E.N. ARNOLD 




Fig. 1 Head and skull of Holaspis. a. Head from above; b. Head from 
side; c. Skull from above; d. Skull from side, fn frontonasal scale, fp 
frontoparietal suture, pa palpebral bone, pm premaxilla, r rostral scale, 
io inferior orbital foramen, so supraocular osteoderms, t triangular scale 
covering area occupied by interparietal and paired frontoparietal scales 
in most other lacertid lizards. 

a full impression of the extent of the dorsoventral flattening, largely 
because in fixed material shrinkage of the jaw muscles pulls the 
kinetic skull into its most retracted position in which the vertical 
extent is greatest (Arnold 1998a). Also, unlike other lacertids, the 
whole of the limbs and tail are depressed in Holaspis. 

Head. The parietal area of the head (Fig. la, b) is flat and 
unarched and the snout is flattened above, being wedge-shaped in 
lateral view. The rostral shield is large, extending far on to the top of 
the snout and contacting the frontonasal scale very broadly. The 
nostrils are placed on the sides of the snout and are set well back 
from its tip. The area of the top of the head usually occupied by the 
interparietal and paired frontoparietal scales in other lacertids is 
covered in Holaspis by a single large triangular scale. The lower 
eyelid has a 'window' composed of enlarged semi-transparent scales. 
In H. guentheri these number 1-5 (mean 3.5, n = 15) while in H. 
laevis there are usually 2-4 (mean 2.3, n = 8) that are sometimes 
black-edged. The scales on the temporal area vary in size: dorsally 
and posteriorly they are typically large and polygonal, whereas 
anteriorly they are much smaller and diagonally elongated, running 
backwards and downwards from behind the eye in irregular lines 
that are separated by somewhat expansible hinge regions. 

The low skull (Fig.lc, d) is more delicately constructed and thin- 
boned than in any other lacertid lizard and the roof of the parietal 
region is so flexible in alcohol-preserved material that it can easily 
be deflected downwards. As in other lacertids, comparatively im- 
mobile sutures in the skull, such as that between the frontal and nasal 
bones, show a considerable overlap between the elements involved, 
giving a measure of rigidity in spite of the thinness of the compo- 
nents. In contrast, the frontoparietal suture, one of the main sites of 






Fig. 2 Scleral ossicles of left eye of lizards, a. Typical lacertid lizard; b. 
Holaspis; c. Platysaurus (Cordylidae). Ossicles are numbered according 
to the system of Gugg (1939); see Underwood (1970). 



cranial kinesis, is a relatively simple abutment without the complex 
interdigitation found at this site in other lacertid lizards. 

The body of the premaxilla is peculiar in forming a broad semicir- 
cular boss that is convex above and supports the extensive rostral 
scale. The nasal openings of the skull are situated posterior to this 
boss and are extremely large. They extend backwards so that the 
primary nasal cavities are broadly exposed dorsally. Of the bony 
elements normally roofing the orbits of lacertid lizards, only the 
palpebral bone is present in its entirety. The usual array of four 
supraocular osteoderms is greatly reduced; the first being absent and 
the others only present in adults, where they are limited to a narrow 
medial fringe. The inferior orbital foramen is very large. Pterygoid 
teeth are absent. The mandibles are slender and shallow and their 
retroarticular processes are directed somewhat ventrally. 

Scleral OSSICLES. In the eye, the scleral ossicles are reduced, 
from the usual lizard number of fourteen that is present in all other 
lacertids, to twelve. This is by the loss of two out of the sequence 
made up of ossicles 5 to 9 (Fig. 2b). The twelve ossicles present are 
so shaped and arranged that the scleral ring is incomplete peripher- 
ally. Instead of extending from the area of the pupil to the vertical 
equator of the eye, the ring is strongly emarginated above and below. 
Dorsally this emargination is produced by the loss of the two 
ossicles, their neighbours extending across the gap so formed and 
overlapping only in the pupillar region. The ventral gap in the outer 
part of the ring is largely a result of the peripheral, radially directed 
part of ossicle 14 being missing but the peripheral sections of 
ossicles 1 and 13 are also skewed away from the gap thus increasing 
its extent. 

BODY. The neck of Holaspis is dorsoventrally flattened, with the 
skin at the sides forming a prominent sharp-edged flap in many 
preserved specimens that is also visible in live animals (Fig. 2, 
Schi0tz, 1960). The flap apparently gains some support from the 
first branchial and hyoid horns of the hyoid apparatus and its edge is 
sometimes marked by a longitudinally oval area of somewhat 
enlarged scales. In other preserved material, in which the pharyngeal 
cavity is expanded dorsoventrally, the flaps are barely apparent, 
suggesting that they are homologous with the slight skin folds which 
occur in this region in many lacertid lizards and which are necessary 
for pharyngeal enlargement. 

The body is strongly depressed and arched in transverse section, 
being convex above and flat beneath. Posterior to the sternally 
connected ribs, the trunk has an elongated oval outline when viewed 
from above and the lateral edges of the body form distinct ridges. 
The dorsal integument consists of two very different types of scaling 
(Fig. 3a). Running along the vertebral region from nape to tail is a 
band of enlarged, broad plates. These are arranged in two longitudi- 
nal series, which are slightly staggered relative to each other. Each 
plate slightly overlaps the one immediately behind it and also, 
medially, the plate diagonally posterior to it in the other row. The 
hinge regions between the plates allow flexibility in the vertical 
plane but do not permit the plates to move much relative to each 
other in the plane of the integument. 

The lateral areas of the dorsal integument are made up of small 
granular scales. At the broadest part of the dorsum, there are 30 to 41 
of these on each side, between the vertebral and ventral plates. These 
small scales often show a differentiation in arrangement between the 
anterior and posterior regions of the back. On the neck and shoul- 
ders, they are non-imbricate and firmly bonded together so that they 
can only move slightly relative to one and other. Further back they 
may gradually alter, so that beyond the sternally connected ribs the 
scales are completely different in character. Here, they are lined up 
in two directions: they are arranged in rows running steeply 



EVOLUTION OF HOLASPIS 



157 





a 



Fig. 3 Dorsal views of the left posterior trunk of Holaspis. a. Skin, 
showing double vertebral band of enlarged scales and small lateral 
scales (only three rows shown), b. Skeleton showing elongated anterior 
free ribs with long cartilaginous tips; the tips usually overlap each other 
to form a continuous edge to the ribs, but in this cleared and stained 
specimen they have become partly separated. 

posteriorly, and to a much lesser extent outwards from the vertebral 
plates, and they also form transverse rows of which there are two or 
three to each vertebral plate. These rows run slightly posteriorly 
from their origin, but further out, they curve a little so that they run 
more or less directly laterally and may turn slightly forwards before 
they reach the perimeter of the dorsum (Fig. 3a). Each scale in a 
transverse row strongly overlaps its neighbour on its medial flank, 
but if the skin is pulled laterally, it stretches easily and extensively, 
so that each scale is separated from its fellows (Fig. 4). It is then 
sometimes apparent that the scales are interconnected by 'bridges'. 
These are often pigmented and not very elastic and apparently 
contain alpha-keratin, as do the scales. They lie slightly below the 
level of the scales themselves and fall into two groups. One series 
runs approximately laterally from the posterior outer border of each 
scale to insert beneath the inner border of its neighbour. The other 
runs from the anterior inner border of every scale and joins it to the 
posterior inner border of the scale which lies in front of it when the 
skin is unstretched. These longitudinal bridges are the only ones 
immediately visible in preserved material, the lateral ones being 
hidden under the imbricating scales. When the skin is not stretched, 
the bridges are slack and slightly folded. The regions between the 
scales and their bridges are made up of soft, extensible skin. 




Fig. 4 Scales from right lateral skin of posterior trunk, a. Skin 
unstretched, scales overlapping medially and longitudinal bridges 
showing, b. Skin stretched showing lateral and longitudinal bridges and 
expanded areas between the scales. The bridge system is not always 
fully apparent. The arrow show direction of stretching. 



presumably consisting largely of beta-keratin. The development of 
this system of bridges shows very considerable variation among 
specimens. 

When the skin is stretched laterally, each scale moves in a 
transverse direction, the excursion made by the outer scales being 
much greater than that made by the inner ones. This results in the 
originally curved transverse rows approximating more closely to a 
straight line. These movements can produce at least 50% increase in 
the area of skin. This ability of the skin to expand is not so highly 
developed as in many snakes, but it is certainly unique among the 
Lacertidae and probably among other lizard groups. 

The dorsolateral skin, when unexpanded, has one or more longi- 
tudinal folds on either side. The whole of the dorsal integument is 
rather loosely attached to the underlying musculature by connective 
tissue, as in other lacertids. The area of granular scales passes round 
the sharp-edged lateral border of the body to contact the ventral 
plates. These are large and arranged in six longitudinal rows, as in 
many other lacertids, and are rectangular showing little imbrication. 
The collar is straight-edged and again not strongly overlapping. 

Holaspis has 25-26 presacral vertebrae in males and 25-27 in 
females. These numbers are unexceptional for lacertids in which the 
majority of species have 25-29 presacrals with extremes of 23 and 
33 and show sexual difference in average number of dorsal verte- 
brae. The vertebrae of Holaspis differ from those of other lacertids 
in being distinctly depressed with virtually no neural crest or spine 

In most lacertid lizards the dorsal ribs can be divided into three 
groups: 1. the thoracic ribs attached to the sternum and xiphister- 
num; 2. the anterior free dorsal ribs which are unattached distally 
and have prominent cartilaginous extensions at their tips; and 3. the 
posterior free dorsal ribs which are usually about two-thirds the 
length of the more anterior ribs and have no cartilaginous exten- 
sions. In Holaspis, there are 7-8 anterior free ribs in males and 8-9 
in females. They are markedly elongate compared with those of 
other lacertids, being considerably longer than the thoracic ribs and 
about twice as long as the posterior free dorsal ones. Their 
cartilaginous extensions are also exceptionally long and are turned 
backwards, each extending beneath the next posterior rib and run- 
ning parallel with its own cartilaginous process (Fig. 3b). These 
overlapping processes are bound together by loose connective tissue 
and form a smooth border to the series of elongated ribs. It is this 
border which forms the prominent edge of the body that runs slightly 
ventrally and backwards to terminate just in front of the anterior 
border of the hind leg. The termination is enclosed in a fold of loose 
skin that connects it to the underside of the thigh. 

The sternum of Holaspis has an extremely large central fontanelle 
that occupies most of its area, and the scapulocoracoid plate has two 
foramina compared with one in other lacertid lizards. 

Holaspis is peculiar among lacertid lizards in having prominent 
slips of the intercostalis scalaris muscle (Maurer 1896) running 
from the tips of the anterior free ribs forwards and somewhat 
inwards to insert on the upper surface of the rectus abdominis 
muscle above the outer edge of the second row of ventral scales. The 
muscle fibres to the more anterior free ribs form a single block but 
those to the more posterior ones comprise separate slips. 

Tail. In nearly all lacertid lizards, the tail is cylindrical and at 
most slightly flattened dorsoventrally at its base. It is covered by 
whorls of numerous subequal scales there being two whorls to each 
caudal vertebra. Deviations from this pattern are usually slight but 
Holaspis differs radically. Its tail (Fig. 5) is somewhat dorsoven- 
trally compressed and above has a double row of broad plates, which 
is a direct continuation of the series on the body. These enlarged 
scales differ from those on the back in being arranged in simple 



158 



E.N. ARNOLD 








Fig. 5 Tail of Holaspis. a. Proximal segment from above; b. Proximal 
segment from below; c. Transverse section. 

lateral pairs instead of being staggered. The plates each have several 
sense organs on their posterior border and, as the usual number of 
sense organs per dorsal caudal scale in other lacertid lizards is one, 
it is likely that the plates have replaced a number of smaller scales. 
The wide double band is flanked by one or two (rarely three) 
longitudinal rows of narrower scales, one row being frequent in H. 
laevis and one or two in H. guentheri. The number of rows some- 
times increases anteriorly and these scales are replaced by granules 
on the tail base. The median part of the ventral surface of the tail is 
formed by another row of wide, paired plates, again replacing 
multiple small scales in other lacertid lizards. 

The lateral edges of the tail are serrated and consist of a single row 
of strongly modified scales. In transverse section, each of these 
scales is more or less triangular, the broad base joining the tail, the 
apex pointing outwards; in this plane, the lateral scales curve 
downwards. Viewed from above, these scales are again approxi- 
mately triangular, the point being directed obliquely backwards. 
Proximally, the longitudinal axis of these scales is parallel with that 
of the whole tail; distally, their anterior edges tend to be twisted 
downwards so that their longitudinal section here runs backwards 
and slightly upwards. Each lateral scale is capable of some move- 
ment since it is connected with contiguous scales in its whorl by 
flexible hinge regions. However, the motion is limited by the scale 
interlocking with its anterior and posterior neighbours. On the 
underside of each of these scales, parallel with and close to the 
trailing edge, is a slit-shaped cavity. The anterior portion of the 
following scale projects into this, giving the lateral fringes consider- 
able stiffness. 

LIMBS. The spans of the fore and hind limbs approach equality 
more closely than in any other lacertid lizard, the index, forelimb 
span/hindlimb span, being 0.85 in males (n = 3), and 0.85 in females, 
(n = 4) while the total range for the Lacertidae is 0.53-0.85 (Arnold 
1998b). 

The forelimbs are rather flattened and the single band of enlarged 
scales, present on the anterior surface of the upper limb of most 
lacertids, occurs in Holaspis too. However, instead of being continued 
as a single band on the lower limb, it is replaced by two parallel ones, 
one dorsal, the other ventral with their zig-zag line of contact forming 
a forwardly directed edge that may sometimes be quite acute. 

The hind leg is similarly markedly depressed and the proximal, 
femoral segment has large plates above and below the leading edge; 



a 



Fig. 6 Right manus of lacertid lizards, a. A form frequently climbing on 
steep open surfaces (Lacerta oxycephala); b. Holaspis guentheri. 

only the lower series reaches the front of the cms. The greater part of 
the trailing area of the hind leg is formed by a web of loose elastic 
skin, which becomes taut when the leg is partly extended. This 
'patagium' is continuous with a series of about four large sharp- 
edged, sometimes interlocking, scales which make up the trailing 
edge of the crus. These scales are generally similar in construction 
and arrangement to those on the lateral edges of the tail 

The manus and pes of Holaspis show strong development of a 
syndrome of features characteristic of lacertid lizards that climb on 
continuous open surfaces (Fig. 6; Arnold, 1998b) and many features 
are better developed than in other forms. The longest digit is number 
4 and all digits are strongly latero-mesially compressed; some digits 
are flexed downwards at the articulation of phalanges 1 and 2 and in 
most there is upward flexure at the penultimate articulation. Phalanges 
are very slender, the penultimate ones being very long and markedly 
curved downwards. In the manus intermediate phalanges of digits 3 
and 4 are shorter than the ones bordering them and the same is true 
of intermediate phalanges of digits 3 and 4 of the pes. The final 
phalanx of each digit and the claw that covers it is short, deep and 
recurved. The large ventral digital tendons are offset from the 
articulations in the regions where digits can be abruptly flexed 
downwards. The articulations within the digits, except the most 
distal, are simple involving single cup and ball arrangement and the 
digits are abruptly flexed in the horizontal plane, both mesially and 
laterally, especially in the area of the penultimate articulation. 

The manus of Holaspis has the following additional derived 
features. Digits 2-5 are subequal in length, and numbers 3 and 4 are 
conjoined for the length of their first phalanx. The shortening and 
downward flexure of phalanx 2 digit 3 and phalanges 2 and 3 of digit 
4 is much more marked than in other lacertid lizards. 

Digits 3, 4 and 5 of the pes each have a lateral fringe of interlock- 
ing triangular scales, which extends distally to the base of the 
penultimate phalange. That on the fifth toe is continuous with the 
similar scales on the trailing edge of the crus. Digits 4 and 5 also 
have a similar mesial fringe. 

COLOUR in life. In life, Holaspis is blue-black with several 
longitudinal pale stripes on the dorsum, the two on the vertebral 
plates being tinged blue posteriorly. The tail has a series of large, 
light, intensely blue spots on its upper surface and its lateral fringes 
are yellow, while the belly is red. 



EVOLUTION OF HOLASPIS 



159 



BEHAVIOUR 



Holaspis guentheri occurs especially in rain-forest situations while 
H. laevis also extends into savannah. These lizards are nearly always 
observed at some height on the trunks and branches of standing 
trees, occurring at least up to 30m, and do not usually come down to 
the ground (H. guentheri: H. Lang in Schmidt. 1919; Perret & 
Mertens. 1957; Schi0tz & Vols0e, 1959; Laurent, 1964; Dunger, 
1967; P. Agland, pers. comm. A. P. Mead, pers. comm. H. laevis: 
Barbour & Loveridge, 1928; Loveridge, 1951, 1953; De Witte, 
1953; Branch, 1998; D. G. Broadley pers. comm.), although they 
can occur on fallen timber (H. Lang in Schmidt, 1919). Holaspis 
spp. are active hunters, constantly moving and searching and often 
investigating crevices (P. Agland, pers. comm.) in which they also 
frequently hide when disturbed and at night (H. Lang in Schmidt, 
1919; Loveridge, 1951; Laurent, 1964; pers. obs. on captive ani- 
mals). They are extremely agile, moving with ease on vertical and 
overhanging surfaces (H. Lang in Schmidt, 1919). Holaspis appear 
to thermoregulate and at times basks in patches of sunlight for at 
least up to ten minutes (Dunger, pers comm.; P. Agland, pers 
comm.). As in many other basking lacertid lizards, the body is 
spread and flattened by the dorsal ribs being rotated forwards and in 
Holaspis the body becomes as flat and round as a coin (Dunger pers. 
comm.. P. Agland pers comm.). 

Holaspis is unique among lacertid lizards in being able to glide 
between trees. This behaviour was first formally reported in H. 
guentheri in Ghana by Schi0tz & Volsoe (1959) and subsequently 
confirmed by P. Agland in Cameroun and A. P. Mead in Nigeria 
(pers. comms). Earlier reports also provide some collaboration. C. J. 
P. Ionides (quoted by Loveridge, 1955) noted that in Tanzania H. 
laevis covers long distances in leaps between trees and Laurent 
(1964) reported that local people in northern Angola said that H. 
guentheri can fly. According to Schi0tz & Vols0e (1959). this lizard 
starts from a head-downwards position, high on a tree trunk from 
which it leaps outwards and glides steeply downwards. The trajec- 
tory later becomes shallower, and just before the lizard alights, it 
turns slightly upwards. For most of the glide, the lizard is orientated 
with its sagittal axis along the direction of motion, but towards the 
end this becomes perpendicular to it, the lizard stalling and reducing 
speed by this means. In one measured leap a lizard travelled 10.5m 
horizontally while dropping 9m, an overall angle of about 42° from 
the horizontal. Holaspis appears capable of selecting a target before 
launching itself, and of changing direction in mid-flight. 

Among the H. guentheri observed by P. Agland (pers. comm.) one 
glided 30m at an angle of 10-20°, another travelled 25m and a third 
6m. Motion was fast and straight and again appeared to be directed. 
In some cases there was an initial drop before the trajectory levelled 
out but in one instance a lizard running horizontally on a branch 
launched itself into the air without much fall before stabilising its 
flight path. At the end of a dive animals again alighted head 
upwards, landing very fast and sometimes immediately running 
upwards. Holaspis clearly has the ability to maintain its belly- 
downwards posture in the air with limbs spread and to change 
orientation as appropriate. 



FUNCTIONAL ANATOMY 



In this section an assessment is made as to whether particular 
morphological apomorphies of Holaspis could have evolved through 
direct adaptation by natural selection in connection with one or more 
of its special behaviours: frequent locomotion on very steep often 



vertical open surfaces, use of very narrow crevices, and gliding. 
Assessment is made on two criteria: 1 . perceived functional benefit 
of the apomorphies in the activities concerned; and 2. whether 
similar apomorphies have appeared independently in other lizards 
that have evolved similar behaviours. The second criterion is most 
convincing if there are multiple independent origins of the apomorphy 
and if these origins are correlated with appearance of the relevant 
behaviour on the lineages of the taxa concerned. Even if there is a 
prima facie case for functional advantage of an apomorphy in 
connection with a particular behaviour, its absence in forms that 
have evolved the behaviour independently raises the possibility that 
it is not connected with the activity concerned. Alternatively, it may 
represent one of several strategies with other taxa gaining similar 
advantages in different ways. 

LOCOMOTION IN tree BOLES. The functional advantages of near- 
equality in fore and hind limb spans, and of characteristic foot 
architecture, in climbing on steep open surfaces has been discussed 
elsewhere (Arnold 1998b). These features are particularly well 
developed in Holaspis and presumably related to the abundance of 
such surfaces in its environment. The unique manus features of 
Holaspis suggest the forelimbs are sometimes used in parasagittal 
planes (Arnold. 1998b). This may be when the lizard launches itself 
from a head-downwards position on a steep surface. Extending the 
forelimbs at this time would push the foreparts of the body out into 
a more horizontal position, putting it closer to its orientation when 
gliding and making an outward leap easier. 

USE OF CREVICES. Features that confer advantages in crevice use 
and the functional basis for this has already been surveyed (Arnold 
1998a). Many derived features of Holaspis occur in other lacertids 
that use rock crevices, having developed independently at least once 
in archaeolacertas (Lacerta spp.), and in Oma.nosa.ura cyanura and 
some populations of Podarcis hispanica. These forms show many 
apomorphies similar to those of Holaspis although the features are 
less developed than in this form, especially the degree of flattening 
of the head, body and limbs. These low vertical dimensions enable 
lizards to enter narrow crevices and a variety of cranial features 
(Appendix 1. numbers 4, 6, 7, 8, 10 and 12) results in a deformable 
skull that can be inserted into irregular spaces. Increased cranial 
kinesis enables the skull to be flattened further by protraction on 
entering a crevice and locked into place by subsequent retraction. As 
a result of flattening of the skull, the eyes, which are large, project 
well above it during normal activities and this potentially impedes 
entry into crevices. However, in lacertids including Holaspis each 
eye is pushed downwards as the lizard enters a crevice by contact 
with the crevice roof so that its lower surface deflects the flexible 
membrane crossing the greatly enlarged inferior orbital foramen. 
This enables the lower part of the eye to project into the buccal 
cavity, so that it can be housed within the depth of the head. 
Reduction of the supraocular osteoderms increases the flexibility of 
the skin over the eyes so that its geometry can alter during their 
depression. Reduced overlap of collar and belly scales increase 
smoothness enabling lizards to move easily both forwards and 
backwards in crevices. Some or all these features are paralleled in 
many non-lacertid crevice users including skinks (such as Mabuya 
laevis and M. sulcata), xantusiids (Xantusia henshawi), geckos 
(Afroedura) and iguanids (Sauromalus, Opiums). 

Holaspis has other features not found in other crevice-using 
lacertids but present in the most extremely flattened exploiters of 
rock crevices in other families, such as Platysaurus (Cordylidae) 
and Tropidurus semitaeniatus (Iguaninidae, Tropidurinae) and prob- 
ably functionally associated with such strong depression. Among 
these is modification of the scleral ossicles, so that there is one or 



160 



E.N. ARNOLD 



more windows in the scleral ring (Fig. 2). These enable the eyeballs 
to distort and flatten, so they can be housed in the narrow space 
available within the head. Other shared features, which also contrib- 
ute to low vertical dimensions, are depression of the body vertebrae 
and reduction of the crests on their neural arches. 

Shortening and consequent decrease in mass of the adductor 
muscles associated with reduction in head height probably plays a 
critical role in facilitating the evolution of this cranial morphol- 
ogy. Curtailed mass reduces the power of the muscles, so a 
particularly strong, thick arched parietal area of the skull is no 
longer necessary to resist their action, and this also permits the 
posterior skull roof to become thin and flat. Similarly the mandi- 
bles are subjected to reduced forces in biting and can consequently 
be more slender with smaller vertical dimensions. However, such 
shortening of the muscles carries penalties in terms of reduced 
efficiency in biting and prey handling (Arnold, 1998a). Change in 
geometry of the skull during the retraction phase of cranial kinesis 
ameliorates this effect by improving their angle of action on the 
jaw and the length of their excursion. This phenomenon is promi- 
nent in Platysaurus, which has an expansion area in the skin on 
the anterior cheek that accommodates the changes involved in the 
substantial kinetic movement. The similarly orientated hinge 
regions of Holaspis, between the small scales found in this region, 
indicates that its skull is similarly highly mobile, as does the 
simplified fronto-parietal suture. 

The downward flexion of the retroarticular process of the mandi- 
ble may permit a longer and more efficient depressor mandibulae, in 
spite of the flattened head, although this feature is not paralleled in 
other very flattened crevice users. Other characteristics of Holaspis 
could also plausibly be considered as adaptations to crevice use, but 
are not repeated in functional analogues. Thus, the large plates along 
the back which might possibly increase smoothness and so ease 
mobility within crevices; although neither Platysurus or T. 
semitaeniatus have this feature. The enlarged sternal fontanelle 
could similarly be thought to increase flexibility in this region, but 
Platysaitrus has no fontanelle at all. 

Another complex of Holaspis features may also be related to 
use of crevices, specifically those beneath bark. This involves the 
snout which is wedge shaped in lateral view (unlike that of rock 
crevice dwellers), with the bizarre flattened boss formed from the 
premaxilla, and nostrils set back from the snout tip and low on its 
sides. Such features are not found even in extremely flattened 
lizards using rock crevices, but they do occur in the flattened 
lygosomine skink Aulacoplax, which habitually conceals itself in 
the narrow interstices between the bases of the fronds of screw 
pines (Pandanus spp.) (Brown and Fehlmann, 1958). This arrange- 
ment may enable the skink to enlarge interstices so they are broad 
enough to take the rest of the animal as it moves forwards. 
Holaspis may possibly do the same when pushing beneath flexible 
pieces of loose bark. The frequent presence of longitudinal 
scratches on the dorsum of the head suggests this may be the case. 
Fusion of the frontoparietal and interparietal scales may increase 
strength and smoothness of the head surface but has no parallels 
elsewhere. 

Gliding. Since Holaspis descends through the air in a controlled 
way at relatively shallow angles it glides rather than parachutes. 
Gliders depend on the possession of an aerofoil which extracts a lift 
component as the animal moves through the air, the lift counteract- 
ing the force of gravity. For gliding at shallow angles to be possible, 
the ratio of surface: body weight needs to be high. Some other 
gliding lizards have a specialised lift surface that provides this. In 
the agamid Draco, this is formed from a membrane supported by the 



elongated abdominal ribs while in the gecko Ptychozoon there are 
flaps attached to the sides of the belly that fold out, increasing 
surface area. In Holaspis, it is the whole body that acts as an aerofoil 
and some features that also confer performance advantage in using 
crevices contribute to its formation. This is particularly true of 
dorsoventral compression, but low ossification of the skull must 
help increase the surface: weight ratio. Other features appear to be 
specifically associated with gliding and are not found in crevice 
dwellers, although they may occur in other gliders. Included here is 
low ossification of the pectoral girdle and perhaps that of the 
sternum and depression of the legs and tail. This last feature, 
together with development of distinct trailing edges on the limbs, 
also occurs in Draco and Ptychozoon, which do not enter very 
narrow crevices. Surface area is further increased by the lateral flaps 
on the neck and the webs of skin that form the trailing edge of the 
proximal hind legs, both features again found in Draco and 
Ptychozoon. The modified scales on the sides of the tail, on the 
trailing edge of the crus and on the hind digits also increase surface 
very efficiently, forming stiff lateral fringe-like extensions with 
little increase in weight. They are exactly paralleled in structure and 
function by scales on the hind side of the thigh and tail base in 
Draco, while Ptychoz.oon has analogous cutaneous extensions along 
the length of the tail. 

Holaspis is able to produce further temporary increase in sur- 
face area by lateral expansion of the abdominal region so that this 
becomes almost disc-like. The increase in surface area is brought 
about by the long free dorsal ribs being rotated forward around 
their articulations with the body vertebrae, so that instead of being 
directed diagonally backwards they project more laterally. In 
Holaspis, the gain in surface area this produces is large because 
the ribs are long. The overlapping flexible rib tips form a continu- 
ous lateral edge to the area supported by the ribs and this maintains 
its continuity and longitudinal orientation in spite of the move- 
ments of the ribs themselves. The rotation of the ribs is 
presumably partly brought about by the intercostal muscles, as 
seems to be the case in Draco (Colbert, 1967). But it is likely that 
the well-developed slips of the m. intercostalis scalaris also play a 
part. As they run somewhat diagonally outwards and backwards 
from the in. rectus abdominis to the rib tips, their contraction 
would also help swing the ribs forwards; at the same time the ribs 
would tend to bow laterally and bend distally downwards. The 
contraction would also raise the m. rectus abdominis and with it 
the ventral integument which is closely attached. These move- 
ments would produce a more aerodynamically efficient transverse 
section in which the dorsal surface was more strongly convex and 
the belly flat or slightly concave. 

The skin must stretch to allow for the increase in lateral area that 
the rib movements produce. Its distinctive structure permits this, for 
expansion occurs not only at the longitudinal lateral skin folds but 
also at the extensible areas between the scales. The bridges that often 
join the scales limit the direction and extent of expansion; they also 
help distribute it evenly throughout the skin, discouraging wrinkling 
and so contributing to a smooth surface. The looseness of the 
connection between the skin and underlying structures usual in 
lizards is also important in allowing skin tension to be evenly 
distributed. 

It is probable that the band of large broad plates in the vertebral 
region also has a function in producing as good an aerofoil as 
possible. As the hinge regions between the plates are virtually 
inelastic, the whole area can be regarded as a single lamina which is 
firmly fixed at the occiput and at the tail base. When such an 
elongate lamina is stretched over a flat or convex surface, and placed 
under tension, it becomes very resistant to lateral deformation. This 



EVOLUTION OF HOLASPIS 



161 



effect is increased when the tension is both along and across the 
lamina. It can be demonstrated by placing a strip of paper on plane 
surface and putting it under longitudinal tension, after which dis- 
placing the intermediate area laterally, even to a small extent, 
becomes very difficult. 

Such rigidity appears to be developed in the vertebral area of 
Holaspis. which extends over the convex dorsum of the body. 
Tension is generated by the lateral skin being pulled outwards during 
rib spreading. This movement distorts the large plates and their 
hinge regions slightly, so that there is a small widening and longitu- 
dinal contraction of the vertebral band. As this is firmly attached at 
the occiput and the tail, tension within it is thereby increased. The 
slight movements of the plates exhaust the very limited internal 
mobility of the band, increasing its lateral rigidity further. These 
processes can be discerned in the detached dorsal skin of an alcohol- 
preserved Holaspis. Lateral tension alone, produces a longitudinal 
contraction of the vertebral band whereas, if it is applied when the 
ends of the band are fixed to the substrate, the band becomes 
laterally rigid. 

The rigid vertebral band ensures that the extended lateral skin is 
spread evenly on both sides of the body, again helping to avoid the 
tendency of the tense skin to wrinkle. It may also act to keep the 
body straight during gliding by restricting lateral bending. This 
effect can be simulated by attaching a strip of adhesive paper tape 
along the side of an elongate rubber balloon. When this is inflated, 
the stretched wall of the balloon exerts tension on the paper strip, 
which represents the vertebral band of Holaspis and the air pressure 
provides support for this in an analagous manner to the body of the 
lizard. A balloon modified in this way is substantially harder to bend 
sideways than an unmodified one. 

When the lateral skin of Holaspis is unstretched, the vertebral 
band is slack and capable of rucking upwards at its hinges. This 
permits the lizard a normal amount of lateral movement, when for 
instance walking rather than gliding, since the band now lacks 
lateral rigidity. 

The surface: weight ratio ( ' wing'-loading) of Holaspis was roughly 
assessed on the assumption the whole animal acts as an aerofoil. 
Area was found by placing straightened preserved lizards belly 
downwards on squared paper and tracing their outline; maximum 
lateral extent of the body was then estimated by comparison with 
photographs of animals basking with their bodies fully expanded, 
and by stretching the lateral skin. Weight was calculated on the 
assumption that live lizards weigh 10% more than alcohol preserved 
ones (Colbert, 1 967) The loadings for four individual adult Holaspis 
varied between 0.26 and 0.37 gm/cm 2 . These are relatively small 
figures when compared with those for Draco (Colbert, 1967). Such 
low loading is probably necessary to compensate for the relatively 
poor general aerodynamic shape of Holaspis. 

Functional significance of other characters. A minority 
of derived features of Holaspis are not functionally associated 
with its main distinctive behaviours The presence of a window in 
the lower eyelid has developed in a wide range of small lizards 
that bask directly in the sun in relatively dry microclimates 
(Arnold, 1973; Greer, 1983). This means that such lizards can 
reduce the extensive water loss associated with these situations by 
closing their eyes but still retain vision to detect predators, pass- 
ing food items etc. In agreement with this explanation, the window 
is better developed in H. laevis which extends into relatively dry 
savannah, than in H. guentheri which appears to be confined to 
forest. Loss of pterygoid teeth in lacertids tends to correlate with 
the general reduction in ossification found in Holaspis and may 
be a concomitant of this. 



EVOLUTION OF HOLASPIS 



The order in which new features develop and the situations in which 
they do so can often be reconstructed by examining states on side 
branches on the lineage of the taxon concerned. This cannot be done 
with many features of Holaspis as they have evolved within its 
exclusive lineage, which by definition lacks side branches, so other 
cues have to be used for these autapomorphies. However exam- 
ination of the relatives of Holaspis does give some information. 
Thus, a degree of climbing is widespread in lacertids as is a modest 
amount of crevice use. This makes it most parsimonious to assume 
these behaviours precede gliding, which is unique to Holaspis. 
These activities and the morphological adaptations associated with 
them are better developed in Holaspis itself. Improvement in climb- 
ing modifications may possibly have begun first, as climbing steep 
tree boles and branches must precede exploiting crevices in them. 

Animal gliders and fliers can be stable or unstable. In stable ones, 
there is a long lift surface behind the centre of gravity. This means 
that, as an animal glides, any tendency to pitch in the sagittal plane 
around the centre of gravity is self-correcting. In pitching, the long 
posterior lift surface will rise or fall, but the air pressure produced by 
forward locomotion will return it and the animal as a whole to its 
original orientation. Unstable fliers with short lift surfaces behind 
the centre of gravity gain in manoeuvrability but do not self-correct 
and so require sophisticated neurological mechanisms to maintain 
appropriate posture in the air, something that is unnecessary in stable 
forms (Smith, 1952). Unsurprisingly, stable forms evolved before 
unstable ones in most of the main groups of flying animals, namely 
insects, pterosaurs and birds, and possibly bats too (Smith, 1952). 

As might be expected from this, Holaspis is a stable glider. The 
centre of gravity of preserved Holaspis appears to be just behind the 
midpoint between the two pairs of legs. There is therefore a consid- 
erable area of lift surface posterior to the centre of gravity, made up 
of the hind body, hind limbs and tail. Experiments were conducted 
with models made out of laminated cardboard and weighted to give 
a wing loading and weight distribution similar to that of Holaspis. 
When gently launched in the appropriate position, these glided well, 
confirming that a glider of the dimensions and shape of Holaspis is 
stable. 

Given inherent stability, gliding ability seems to require only an 
aerofoil and ability to reach and maintain an appropriate belly-down 
posture with limbs slightly raised, as well as some ability to trim, at 
least initially. In tree frogs Cott, (1926) found adoption of initial 
posture very critical: Phrynohyas venulosa spread its limbs and 
glided when dropped whereas Hyla arborea, which is morphologi- 
cally similar, dropped vertically with legs flailing. It might be 
expected that the ability to adopt the appropriate posture would be 
confined to Holaspis among lacertids as it is the only known glider, 
but when tests were carried out on a number of lacertids this 
propensity was found to be widespread, being present in completely 
terrestrial lizards such as Lacerta agilis and Acanthodactylus 
erythrurus, as well as climbing ones (Arnold, 1989a). The wide 
distribution of this ability suggests it confers advantage in another 
more general context and may have arisen there. This context may 
be terrestrial locomotion. Certainly running lacertids seem to have 
to continuously adjust their body positions and, at some points in the 
stride cycle, they may be balanced on only a single toe (Arnold 
1998b), so good neurological control of posture seems to be essen- 
tial in this activity. 

The production of an aerofoil is likely to largely result from direct 
adaptation, as many of the features of Holaspis appear to confer 
advantage only in gliding. However rib spreading, like balance, is 



162 



E.N. ARNOLD 



coopted from an earlier activity. All lacertids. including Holaspis, 
appeal - to spread their ribs when basking in relatively cool conditi- 
ons, increasing surface area and rate of heat intake. 

Permanent depression of the head, body and limbs of course also 
contributes to the aerofoil and we can ask whether this is a special 
feature of gliding or whether it is coopted from crevice use. As 
already noted there is some phylogenetic evidence of its earlier 
origin for crevice use, something that has occurred in many inde- 
pendent lineages. 

Modifications for gliding in Holaspis are quite extensive, several 
organ systems being involved. It is therefore rather surprising that 
Holaspis has not developed a more efficient aerofoil such as occurs 
in Draco. However to do this would probably involve the develop- 
ment of a delicate patagium or extensive lateral skin flaps. It is likely 
that such structures would interfere with the lizard's ability to enter 
and move in the narrow crevices it regularly utilises. Consequently 
Holaspis is restricted to using means of broadening the body that do 
not project exteriorly. 

As a stable glider, Holaspis is very dependent on its long tail, but 
this can break easily, even close to its base where loss might make it 
unstable in the air and reduce its ability to control its glide path. 
Nonetheless the tail is lost and often regenerated in many individuals 
(Arnold, 1984). This emphasises the importance of tail loss as an 
antipredator device and suggests the cost: benefit ratio still favours 
frequent tail loss in Holaspis even though locomotory costs may 
well be high. 

It might be thought that cases like Holaspis, where entrance into 
a new life mode has been dependent on multiple exaptations, are 
rare. But this phenomenon occurs in another instance where aerial 
locomotion has been attained, that of birds where feathers and the 
complex mechanism of wing folding arose long before gliding or 
active flight (Gauthier & Gall, 2001). 



ACKNOWLEDGEMENTS. I am grateful to P. Agland, D. Broadley, G. Dunger, 
the late A. Loveridge, and A. P. Mead for information about the behaviour of 
Holaspis in the field, and to Garth Underwood for helpful discussion. An 
earlier version of this paper formed part of a D. Phil thesis submitted to the 
University of Oxford. In connection with this 1 thank my supervisors, the late 
A. J. Cain and the late N. Tinbergen, and the Scientific Research Council of 
the United Kingdom for providing funding. 1963-1966. 



REFERENCES 



Arnold, E. N. 1973. Relationships of the Palaearctic lizards assigned to the genera 

Lacerta, Algyroides and Psammodromus (Reptilia: Lacertidae). Bulletin of the 

British Museum (Natural History) 25: 289-366. 
1984. Evolutionary aspects of tail shedding in lizards and their relatives. Journal 

of Natural History 18: 127-169 
1988. Caudal autotomy as a defense. In C. Gans & R. B. Huey (eds) Biology of the 

Reptilia 16B: Defense and Life Histoiy. Alan R. Liss, New York. 
1989a. Systematics and adaptive radiation of Equatorial African lizards assigned 

to the genera Adolfus. Bedriugaia. Gastropholis, Holaspis and Lacerta (Reptilia: 

Lacertidae). Journal of Natural History 23: 525-555. 
1989b. Towards a phylogeny and biogeography of the Lacertidae: relationships 

within and an Old-World family of lizards derived from morphology. Bulletin of the 

Natural History Museum London (Zoology) 55: 209-257. 
1990. Why do morphological phylogenies vary in quality? An investigation based 

on the comparative history of lizard clades. Proceedings of the Royal Society of 



London B 240: 135-172. 
1991 . Relationship of the South African lizards assigned to Aporosaura, Meroles 

and Pedioplanis (Reptilia: Lacertidae). Journal of Natural History 25: 783-807. 
1994. Investigating the origins of performance advantage: adaptation, exaptation 

and lineage effects. In P. Eggleton & R. Vane-Wright (eds) Phylogenetics and 

Ecology. Linnean Society of London and Academic Press, London. Pp. 123-168. 

1998a. Cranial kinesis in lizards. Variations, uses and origins. Evolutionary 

Biology 30: 323-357. 

1998b. Structural niche, limb morphology and locomotion in lacertid lizards 

(Squamata, Lacertidae); a preliminary survey. Bulletin of the Natural Histoiy Mu- 
seum London (Zoology) 64: 63-89. 

Barbour, T. & Loveridge, A. 1 928. A comparative study of the herpetological faunae 

of the Uluguru and Usumbara Mountains. Tanganyika Territory with descriptions of 

new species. Memoirs of the Museum of Comparative Zoology 50: 84-265. 
Branch, W. R. 1998. Field Guide to Snakes and Other Reptiles of Southern Africa. 

Struik. Cape Town. 
Broadley, D. 2000. Lacertidae. Holaspis laevis (Werner. 1895). Eastern serrate-toed 

tree lizard. African Herp News 31: 13-14 
Brown, W. C. & Fehlmann, A. 1 958. A new genus and species of arboreal lizards from 

the Palau islands. Occasional Papers. Natural Histoiy Museum. Stanford University 

6: 1-7. 
Colbert, E. H. 1967. Adaptations for gliding in the lizard Draco. American Museum 

Novitates 2283: 1-20. 
Cott, H. B. 1926. Observations on the life-habits of some batrachians and reptiles from 

the lower Amazon: and a note on some mammals from Marajo island. Proceedings 

of the Zoological Society of London 1926: 1159-1178. 
Darwin, C. 1872. On the Origin of Species. 6"' edition. John Murray. London. 
De Witte, G. F. 1953. Exploration du Pare National de I'Upemba 6: Reptiles. Brussels. 
Dunger, G. T. 1967. The lizards and snakes of Nigeria, part 2: The Lacertids of Nigeria. 

Nigerian Field 32: 117-130. 
Gauthier, J. & Gall, L. F. (eds) 2001. New Perspectives on the Origin and Early 

Evolution of Birds. Peabody Museum, New Haven. Connecticut. 
Gould, S. J. 2002. The Structure of Evolutionary Theory. Belknap, Harvard. 
& Vrba, E. 1982. Exaptation -a missing term in the science of form. Paleobiology 

8: 4-15. 
Greer, A. E. 1983. On the adaptive significance of the reptilian spectacle: the evidence 

from scincid. teiid and lacertid lizards. In: A. G. J. Rhodin & K. Miyata (eds). 

Advances in Herpetology, Essays in Honor of Ernest E. Williams. Museum of 

Comparative Zoology. Cambridge. Massachusetts. Pp.2 13-221. 
Gugg, W. 1939. Der Skleralring der plagiotremen Reptilien. Zoologische Jahrbiicher, 

Abteilung fiir Anatomic 65: 339— H 6. 
Harris, D. J., Arnold, E. N. & Thomas, R. H. 1998. Relationships of lacertid lizards 

(Reptilia: Lacertidae) estimated from mitochondrial DNA sequences and morphol- 
ogy. Proceedings of the Royal Society of London B 265: 1939-1948. 
Laurent, R. F. 1964. Reptiles et amphibiens de 1' Angola. Companhia de Diamantes de 

Angola: Publicacoes Culturais 67: 1-165. 
Loveridge, A. 195 1 . On reptiles and amphibians from Tanganyika collected by C. J. P. 

Ionides. Bulletin of the Museum of Comparative Zoology. Hanard 106: 175-204. 

1953. Zoological results of a fifth expedition to East Africa III: Reptiles from 

Nyasaland and Tete. Bulletin of the Museum of Comparative Zoology. Hanard 110: 
143-487. 

1955. On a second collection of reptiles and amphibians taken in Tanganyika 

Territory by C. J. P. Ionides. Journal of the East Africa Natural Histoiy Society 22: 
168-198. 

Maurer,F. 1896. Die ventraleRumpfmuskulatureiniger Reptilien. Einervergleichender- 

anatomische Untersuchungen. Festschrift fiir Gegenbaur I: 183-256. 
Perret, J. L. & Mertens, R. 1957. Etude d'une collection herpetologique faite au 

Cameroun de 1952 a 1955. Bulletin de I'Institut Francois d'Afrique Noire 19: 548- 

601. 
Schietz, A. 1960. Vestafrikas flyvende firben. Naturens Verden. July 1960. 
& Vols0e, H. 1 959. The gliding flight of Holaspis guentheri Gray, a West African 

lacertid. Copeia 1959: 259-260. 
Schmidt, K. P. 1919. Contributions to the herpetology of the Belgian Congo based on 

the collection of the American Congo Expedition 1909-1915. Part I: Turtles. 

crocodiles, lizards and chameleons; with field notes by Herbert Lang and James P. 

Chapin. Bulletin of the American Museum of Natural Histoiy 39: 385-624. 
Smith, J. Maynard. 1952. Importance of the nervous system in evolution of animal 

flight. Evolution 6: 1 27. 
Underwood, G. L. 1970. The eye. In Gans, C. & Parsons. T. S. (eds). Biology of the 

Reptilia 2: 1-97. 



EVOLUTION OF HOLASP1S 

Appendix 1 

Derived features of Holaspis not found in immediate relatives in the 
paraphyletic genus Adolf us. Most features are unique in the Lacertidae 
as a whole and these are denoted by *. Separate postorbital and 
postfrontal bones are primitive in the Lacertidae but fusion is the 
usual condition in the African Equatorial group and separation in 
Holaspis is a reversal (Arnold, 1989a). Most features appear to 
confer a performance advantage in one or more of the characteristic 
behaviours of Holaspis or ameliorate a problem associated with 
them. The behaviours concerned are designated as follows: L - 
locomotion on steep surfaces, C - use of crevices, G - gliding. 
Brackets indicate a relatively minor role. 



163 



Proportions 

1 . Head, body, limbs and tail extremely depressed* 

2. Snout sharply wedge-shaped in lateral view* 

3. Fore and hind limbs approach equality in length* 

Skeleton and musculature 

4. Skull light and very thin-boned with some 
deformability* 

5. Premaxilla forming a large semicircular boss* 

6. Nasal openings of skull very large and widely 
expose primary nasal chambers 

7. Fronto-parietal suture a simple abuttment, not strongly 
interdigitated*. 

8. Postorbital and postfrontal bones separate 

9. Supraocular osteoderms very reduced* 

10. Inferior orbital foramen extremely large 

1 1 . Pterygoid teeth absent 

12. Increased cranial kinesis 

13. Ring of scleral ossicles reduced to twelve and 
emarginated above and below* 

14. Retroarticular process of mandible directed 
somewhat ventrally* 

15. Dorsal vertebrae depressed with very reduced neural 
spines* 

16. Anterior free ribs elongated with long cartilagenous 
extensions at their tips* 

17. Coracoid plate with an extra fontanelle* 

18. Sternal fontanelle very large* 

19. M. intercostalis scalaris well developed, consisting of 
slips originating on tips of anterior free dorsal ribs 



CG 
C 
L 



CG 



20. 



21. 



22. 
23. 



and running forwards and medially to insert on upper 

surface of m. rectus abdominis, above outer margins 

of second row of ventral scales* 

Manus and pes have syndrome of features associated 

with climbing on continuous open surfaces very 

pronounced*. 

Length and downward curvature of penultimate 

phalanges of digits better developed than in other 

lacertids* 

Digits 2-5 of manus subequal in length* 

In manus, shortening and downward flexure of phalanx 

2 of toe, 3 and phalanges 2 and 3 of toe 4, very 

pronounced* 



c 




•(G) 


29. 


C 


30. 




31. 


c 




'(G) 


32. 


C 






33. 


c 




c 


34. 




35. 


c 




c 


36 




37. 


G 


38. 


G 


39. 


CG 


40. 


CG 


41. 


G 





External features 

24. Rostral scale very large, extending on to top of snout C 
with broad frontonasal contact* 

25. Nostrils set back, on sides of snout* C 

26. Interparietal and paired frontoparietal scales all C 
replaced by a single triangular scale* 

27. A window in the lower eyelid made up of 1-5 enlarged 
semitransparent scales 

28. Temporal scales differentiated with anterior ones C 
arranged in diagonal lines* 

Neck with sharp-edged flap on each side when G 

pharynx not expanded* 

Cross section of body convex above and flat below G 

A double series of very large flat scales along (C)G 

vertebral region of body* 

Dorsal scales on sides of posterior trunk laterally G 

expandable* 

Lateral dorsal scales on posterior trunk sometimes G 

joined by a system of bridges* 

Collar and belly scales with very reduced imbrication C 

Tail with two longitudinal rows of broad enlarged G 

scales above and below, the former with multiple 
sensory pores* 

Tail with lateral fringes of interconnected pointed scales* G 
Two rows of large scales on front of forelimbs* CG 

A patagium behind the knee* G 

A row of flat triangular scales on trailing edge of cms* G 
Second and third fingers of manus conjoined at base* L 

Digits 3-5 of pes with a fringe of interlocking G 

pointed scales* 

42. Distinctive colouring* 



GUIDE FOR AUTHORS 



Aims and scope. The Bulletin of the British Museum (Natural 
History) Zoology, was established specifically to accommodate 
manuscripts relevant to the Collections in the Department of Zool- 
ogy. It provides an outlet for the publication of taxonomic papers 
which, because of their length, prove difficult to publish elsewhere. 
Preference is given to original contributions in English whose 
contents are based on the Collections, or the description of speci- 
mens which are being donated to enhance them. Acceptance of 
manuscripts is at the discretion of the Editor, on the understanding 
that they have not been submitted or published elsewhere and 
become the copyright of the Trustees of the Natural History Mu- 
seum. All submissions will be reviewed by at least two referees. 

Submission of manuscripts. Initially three clear, complete cop- 
ies should be submitted in the style and format of the Bulletin. The 
text must be typed double-spaced throughout, including references, 
tables and legends to figures, on one side of A4 paper with 2.5 cm 
margins. All pages should be numbered consecutively, beginning 
with the title page as p. 1 . SI units should be used where appropriate. 

Whenever possible a copy of the text, once the paper has been 
accepted, should also be provided on floppy disc (see below). Discs 
should only be sent after final acceptance, as papers generally need 
revision after refereeing. If it is impossible to provide an appropriate 
disc please ensure that the final typescript is clearly printed. 

Authors are requested to ensure that their manuscripts arc in final 
format, because corrections at proof stage may be charged to the 
author. Additions at proof stage will not normally be allowed. Page 
proofs only will be sent. 

Word-processor discs. Please follow these instructions. 

1 . Ensure that the disc you send contains only the final version of 
the paper and is identical to the typescript. 

2. Label the disc with the author's name, title of the paper and 
the word-processor programme used. Indicate whether IBM or 
Apple Mac (IBM preferred). 

3. Supply the file in the word-processor format; if there is a 
facility to save in ASCII please submit the file in ASCII as well. 

4. Specify any unusual non-keyboard characters on the front 
page of the hard copy. 

5. Do not right-justify the text. 

6. Do not set a left-hand margin. 

7. Make sure you distinguish numerals from letters, e.g. zero (0) 
from O; one ( 1 ) from 1 (el) and I. 

8. Distinguish hyphen, en rule (longer than a hyphen, used 
without a space at each end to signify 'and' or 'to' .e.g. the Harrison- 
Nelson technique, 91-95%, and increasingly used with a space at 
each end parenthetically), and em rule (longer than an en rule, used 
with a space at each end parenthetically) by: hyphen, two hyphens 
and three hyphens, respectively. Be consistent with rule used paren- 
thetically. 

9. Use two carriage returns to indicate beginnings of paragraphs. 

1 0. Be consistent with the presentation of each grade of heading 
(see Text below). 

Title. The title page should be arranged with the full title; name(s) 
of author(s) without academic titles; institutional address(es); sug- 
gested running title; address for correspondence. 

Synopsis. Each paper should have an abstract not exceeding 200 
words. This should summarise the main results and conclusions of 
the study, together with such other information to make it suitable 
for publication in abstracting journals without change. References 
must not be included in the abstract. 

Text. All papers should have an Introduction, Acknowledge- 
ments (where applicable) and References; Materials and Methods 
should be included unless inappropriate. Other major headings are 
left to the author's discretion and the requirements of the paper, 
subject to the Editors' approval. Three levels of text headings and 



sub-headings should be followed. All should be ranged left and be in 
upper and lower case. Supra-generic systematic headings only 
should be in capitals; generic and specific names are to be in italics, 
underlined. Authorities for species names should be cited only in the 
first instance. Footnotes should be avoided if at all possible. 

References. References should be listed alphabetically. Authori- 
ties for species names should not be included under References, 
unless clarification is relevant. The author's name, in bold and lower 
case except for the initial letter, should immediately be followed by 
the date after a single space. Where an author is listed more than 
once, the second and subsequent entries should be denoted by a long 
dash. These entries should be in date order. Joint authorship papers 
follow the entries for the first author and an '&' should be used 
instead of 'and' to connect joint authors. Journal titles should be 
entered in full. Examples: (i) Journals: England, K..W. 1987. Certain 
Actinaria (Cnidaria, Anthozoa) from the Red Sea and tropical Indo- 
Pacific Ocean. Bulletin of the British Museum (Natural History), 
Zoology 53: 206-292. (ii) Books: Jeon. K.W. 1973. The Biology of 
Amoeba. 628 p. Academic Press. New York & London, (iii) Articles 
from books: Hartman, W.D. 1 98 1 . Form and distribution of silica in 
sponges, pp. 453-493. In: Simpson, T.L. & Volcani, B.E. (eds) 
Silicon and Siliceous Structures in Biological Systems. Springer- 
Verlag, New York. 

Tables. Each table should be typed on a separate sheet designed to 
extend across a single or double column width of a Journal page. It 
should have a brief specific title, be self-explanatory and be supple- 
mentary to the text. Limited space in the Journal means that only 
modest listing of primary data may be accepted. Lengthy material, 
such as non-essential locality lists, tables of measurements or details 
of mathematical derivations should be deposited in the Biological 
Data Collection of the Department of Library Services, The Natural 
History Museum, and reference should be made to them in the text. 

Illustrations 

drawings - Figures should be designed to go across single (84 mm 
wide) or double ( 174 mm wide) column width of the Journal page, 
type area 235 x 1 74 mm. Drawings should be in black on white stiff 
card with a line weight and lettering suitable for the same reduction 
throughout, ideally not more than 40' ; . After reduction the smallest 
lettering should be not less than 10pt(3 mm). Tracing paper should 
ideally be avoided because of the possibility of shadows when 
scanned. All artwork must have bulletin, author and figure number 
included, outside of the image area, and must be free of pencil, glue 
or tape marks. 

photographs - All photographs should be prepared to the final size 
of reproduction, mounted upon stiff card and labelled with press-on 
lettering (eg Letraset). They can be mounted on white or black 
background: a black background must be evenly black all over; any 
background must be free of all pencil and glue marks within the 
image area. All figures should be numbered consecutively as a 
single series. Legends, brief and precise, must indicate scale and 
explain symbols and letters. Photos, when components of figure- 
plates should be abutted, trimmed as regular rectangles or close 
trimmed up to edge of specimen. Joins etc. can be removed at the 
scanning stage but at extra cost. Cropping instructions, if any, should 
be indicated on an overlay or marked on a photocopy of the figure. 
SIZE - Maximum size of artwork for use of flatbed scanners is A3. 
Larger artwork has to be reduced photographically prior to scan- 
ning, therefore adding to expense. 

Symbols in text. Male and female symbols within the text should 
be flagged within curly brackets to enable setter to do a swift global 
search. 

Reprints. 25 reprints will be provided free of charge per paper. 
Orders for additional reprints can be submitted to the publisher on 
the form provided with the proofs. Later orders cannot be accepted. 



Editorial: Garth Underwood - Dedication 
51 Hemipenial variation in the African snake genus Crotaphopeltis Fitzinger, 1843 (Serpentes, 

Colubridae, Boiginae) 

T. Ziegler and J. B. Rasmussen 
57 Review of the Dispholidini, with the description of a new genus and species from Tanzania 

(Serpentes, Colubridae) 

D. G. Broadley and V. Wallach 

75 On the African leopard whip snake, Psammophis leopardinus Bocage, 1887 (Colubridae, 

Ophidia), with the description of a new species from Zambia 

B. Hughes and E. Wade 
83 Morphological variation and the definition of species in the snake genus Tropidophis 

(Serpentes, Tropidophiidae) 

S. B. Hedges 
91 Atractaspis (Serpentes, Atractaspididae) the burrowing asp; a multidisciplinary minireview 

E. Kochva 

101 Origin and phylogenetic position of the Lesser Antillean species of Bothrops (Serpentes, 

Viperidae): biogeographical and medical implications 

W. Wuster, R. S. Thorpe, M. da Graga Salomao, L. Thomas, G. Puorto, R. D. G. Theakston and 

D. A. Warrell 
1 07 A contribution to the svstematics of two commonly confused pitvipers from the Sunda 



ion of the geckos: a 21" century appreciation 

nae (Reptilia. Serpentes). with special reference to the otico 



squamate Mesoleptos and the origin of snakes 
ilon 

:aecilians (Amphibia, Gymnophiona) and its systematic utility 
son 

1 55 Holaspis. a lizard that glided by accident: mosaics of cooption and adaptation In a tropical 
forest lacertid (Reptilia. Lacertidae) 



113 


Underwood s 


>if ic 




A P 




123 


The skull of the Ur 
oc< 'pii.ll region 


ope 




1H / 


<ihe 


131 


The Cretaceous m 

■id J i 
Phallus morpholoc 

D J 





Cambridge 

UNIVERSITY PHI 



0968-0470(2002 1 1 )68 :2 ; 1 -2 



ZOOLOGY SERIES 

Vol. 68, No. 2, November 2002